fix: 临时移除context-1m-2025-08-07以确保避免sonnet1m触发429

feat: add Cache TTL Override per account

.github/workflows/security-scan.yml

@@ -32,7 +32,7 @@ jobs:
         working-directory: backend
         run: |
           go install github.com/securego/gosec/v2/cmd/gosec@latest
-          gosec -severity high -confidence high ./...
+          gosec -conf .gosec.json -severity high -confidence high ./...
 
   frontend-security:
     runs-on: ubuntu-latest

backend/.gosec.json

@@ -0,0 +1,5 @@
+{
+  "global": {
+    "exclude": "G704"
+  }
+}

backend/cmd/server/VERSION

@@ -1 +1 @@
-0.1.76
+0.1.83

backend/ent/migrate/schema.go

@@ -650,6 +650,7 @@ var (
 		{Name: "ip_address", Type: field.TypeString, Nullable: true, Size: 45},
 		{Name: "image_count", Type: field.TypeInt, Default: 0},
 		{Name: "image_size", Type: field.TypeString, Nullable: true, Size: 10},
+		{Name: "cache_ttl_overridden", Type: field.TypeBool, Default: false},
 		{Name: "created_at", Type: field.TypeTime, SchemaType: map[string]string{"postgres": "timestamptz"}},
 		{Name: "api_key_id", Type: field.TypeInt64},
 		{Name: "account_id", Type: field.TypeInt64},
@@ -665,31 +666,31 @@ var (
 		ForeignKeys: []*schema.ForeignKey{
 			{
 				Symbol:     "usage_logs_api_keys_usage_logs",
-				Columns:    []*schema.Column{UsageLogsColumns[26]},
+				Columns:    []*schema.Column{UsageLogsColumns[27]},
 				RefColumns: []*schema.Column{APIKeysColumns[0]},
 				OnDelete:   schema.NoAction,
 			},
 			{
 				Symbol:     "usage_logs_accounts_usage_logs",
-				Columns:    []*schema.Column{UsageLogsColumns[27]},
+				Columns:    []*schema.Column{UsageLogsColumns[28]},
 				RefColumns: []*schema.Column{AccountsColumns[0]},
 				OnDelete:   schema.NoAction,
 			},
 			{
 				Symbol:     "usage_logs_groups_usage_logs",
-				Columns:    []*schema.Column{UsageLogsColumns[28]},
+				Columns:    []*schema.Column{UsageLogsColumns[29]},
 				RefColumns: []*schema.Column{GroupsColumns[0]},
 				OnDelete:   schema.SetNull,
 			},
 			{
 				Symbol:     "usage_logs_users_usage_logs",
-				Columns:    []*schema.Column{UsageLogsColumns[29]},
+				Columns:    []*schema.Column{UsageLogsColumns[30]},
 				RefColumns: []*schema.Column{UsersColumns[0]},
 				OnDelete:   schema.NoAction,
 			},
 			{
 				Symbol:     "usage_logs_user_subscriptions_usage_logs",
-				Columns:    []*schema.Column{UsageLogsColumns[30]},
+				Columns:    []*schema.Column{UsageLogsColumns[31]},
 				RefColumns: []*schema.Column{UserSubscriptionsColumns[0]},
 				OnDelete:   schema.SetNull,
 			},
@@ -698,32 +699,32 @@ var (
 			{
 				Name:    "usagelog_user_id",
 				Unique:  false,
-				Columns: []*schema.Column{UsageLogsColumns[29]},
+				Columns: []*schema.Column{UsageLogsColumns[30]},
 			},
 			{
 				Name:    "usagelog_api_key_id",
 				Unique:  false,
-				Columns: []*schema.Column{UsageLogsColumns[26]},
+				Columns: []*schema.Column{UsageLogsColumns[27]},
 			},
 			{
 				Name:    "usagelog_account_id",
 				Unique:  false,
-				Columns: []*schema.Column{UsageLogsColumns[27]},
+				Columns: []*schema.Column{UsageLogsColumns[28]},
 			},
 			{
 				Name:    "usagelog_group_id",
 				Unique:  false,
-				Columns: []*schema.Column{UsageLogsColumns[28]},
+				Columns: []*schema.Column{UsageLogsColumns[29]},
 			},
 			{
 				Name:    "usagelog_subscription_id",
 				Unique:  false,
-				Columns: []*schema.Column{UsageLogsColumns[30]},
+				Columns: []*schema.Column{UsageLogsColumns[31]},
 			},
 			{
 				Name:    "usagelog_created_at",
 				Unique:  false,
-				Columns: []*schema.Column{UsageLogsColumns[25]},
+				Columns: []*schema.Column{UsageLogsColumns[26]},
 			},
 			{
 				Name:    "usagelog_model",
@@ -738,12 +739,12 @@ var (
 			{
 				Name:    "usagelog_user_id_created_at",
 				Unique:  false,
-				Columns: []*schema.Column{UsageLogsColumns[29], UsageLogsColumns[25]},
+				Columns: []*schema.Column{UsageLogsColumns[30], UsageLogsColumns[26]},
 			},
 			{
 				Name:    "usagelog_api_key_id_created_at",
 				Unique:  false,
-				Columns: []*schema.Column{UsageLogsColumns[26], UsageLogsColumns[25]},
+				Columns: []*schema.Column{UsageLogsColumns[27], UsageLogsColumns[26]},
 			},
 		},
 	}

backend/ent/mutation.go

@@ -15061,6 +15061,7 @@ type UsageLogMutation struct {
 	image_count                 *int
 	addimage_count              *int
 	image_size                  *string
+	cache_ttl_overridden        *bool
 	created_at                  *time.Time
 	clearedFields               map[string]struct{}
 	user                        *int64
@@ -16687,6 +16688,42 @@ func (m *UsageLogMutation) ResetImageSize() {
 	delete(m.clearedFields, usagelog.FieldImageSize)
 }
 
+// SetCacheTTLOverridden sets the "cache_ttl_overridden" field.
+func (m *UsageLogMutation) SetCacheTTLOverridden(b bool) {
+	m.cache_ttl_overridden = &b
+}
+
+// CacheTTLOverridden returns the value of the "cache_ttl_overridden" field in the mutation.
+func (m *UsageLogMutation) CacheTTLOverridden() (r bool, exists bool) {
+	v := m.cache_ttl_overridden
+	if v == nil {
+		return
+	}
+	return *v, true
+}
+
+// OldCacheTTLOverridden returns the old "cache_ttl_overridden" field's value of the UsageLog entity.
+// If the UsageLog object wasn't provided to the builder, the object is fetched from the database.
+// An error is returned if the mutation operation is not UpdateOne, or the database query fails.
+func (m *UsageLogMutation) OldCacheTTLOverridden(ctx context.Context) (v bool, err error) {
+	if !m.op.Is(OpUpdateOne) {
+		return v, errors.New("OldCacheTTLOverridden is only allowed on UpdateOne operations")
+	}
+	if m.id == nil || m.oldValue == nil {
+		return v, errors.New("OldCacheTTLOverridden requires an ID field in the mutation")
+	}
+	oldValue, err := m.oldValue(ctx)
+	if err != nil {
+		return v, fmt.Errorf("querying old value for OldCacheTTLOverridden: %w", err)
+	}
+	return oldValue.CacheTTLOverridden, nil
+}
+
+// ResetCacheTTLOverridden resets all changes to the "cache_ttl_overridden" field.
+func (m *UsageLogMutation) ResetCacheTTLOverridden() {
+	m.cache_ttl_overridden = nil
+}
+
 // SetCreatedAt sets the "created_at" field.
 func (m *UsageLogMutation) SetCreatedAt(t time.Time) {
 	m.created_at = &t
@@ -16892,7 +16929,7 @@ func (m *UsageLogMutation) Type() string {
 // order to get all numeric fields that were incremented/decremented, call
 // AddedFields().
 func (m *UsageLogMutation) Fields() []string {
-	fields := make([]string, 0, 30)
+	fields := make([]string, 0, 31)
 	if m.user != nil {
 		fields = append(fields, usagelog.FieldUserID)
 	}
@@ -16980,6 +17017,9 @@ func (m *UsageLogMutation) Fields() []string {
 	if m.image_size != nil {
 		fields = append(fields, usagelog.FieldImageSize)
 	}
+	if m.cache_ttl_overridden != nil {
+		fields = append(fields, usagelog.FieldCacheTTLOverridden)
+	}
 	if m.created_at != nil {
 		fields = append(fields, usagelog.FieldCreatedAt)
 	}
@@ -17049,6 +17089,8 @@ func (m *UsageLogMutation) Field(name string) (ent.Value, bool) {
 		return m.ImageCount()
 	case usagelog.FieldImageSize:
 		return m.ImageSize()
+	case usagelog.FieldCacheTTLOverridden:
+		return m.CacheTTLOverridden()
 	case usagelog.FieldCreatedAt:
 		return m.CreatedAt()
 	}
@@ -17118,6 +17160,8 @@ func (m *UsageLogMutation) OldField(ctx context.Context, name string) (ent.Value
 		return m.OldImageCount(ctx)
 	case usagelog.FieldImageSize:
 		return m.OldImageSize(ctx)
+	case usagelog.FieldCacheTTLOverridden:
+		return m.OldCacheTTLOverridden(ctx)
 	case usagelog.FieldCreatedAt:
 		return m.OldCreatedAt(ctx)
 	}
@@ -17332,6 +17376,13 @@ func (m *UsageLogMutation) SetField(name string, value ent.Value) error {
 		}
 		m.SetImageSize(v)
 		return nil
+	case usagelog.FieldCacheTTLOverridden:
+		v, ok := value.(bool)
+		if !ok {
+			return fmt.Errorf("unexpected type %T for field %s", value, name)
+		}
+		m.SetCacheTTLOverridden(v)
+		return nil
 	case usagelog.FieldCreatedAt:
 		v, ok := value.(time.Time)
 		if !ok {
@@ -17745,6 +17796,9 @@ func (m *UsageLogMutation) ResetField(name string) error {
 	case usagelog.FieldImageSize:
 		m.ResetImageSize()
 		return nil
+	case usagelog.FieldCacheTTLOverridden:
+		m.ResetCacheTTLOverridden()
+		return nil
 	case usagelog.FieldCreatedAt:
 		m.ResetCreatedAt()
 		return nil

backend/ent/runtime/runtime.go

@@ -779,8 +779,12 @@ func init() {
 	usagelogDescImageSize := usagelogFields[28].Descriptor()
 	// usagelog.ImageSizeValidator is a validator for the "image_size" field. It is called by the builders before save.
 	usagelog.ImageSizeValidator = usagelogDescImageSize.Validators[0].(func(string) error)
+	// usagelogDescCacheTTLOverridden is the schema descriptor for cache_ttl_overridden field.
+	usagelogDescCacheTTLOverridden := usagelogFields[29].Descriptor()
+	// usagelog.DefaultCacheTTLOverridden holds the default value on creation for the cache_ttl_overridden field.
+	usagelog.DefaultCacheTTLOverridden = usagelogDescCacheTTLOverridden.Default.(bool)
 	// usagelogDescCreatedAt is the schema descriptor for created_at field.
-	usagelogDescCreatedAt := usagelogFields[29].Descriptor()
+	usagelogDescCreatedAt := usagelogFields[30].Descriptor()
 	// usagelog.DefaultCreatedAt holds the default value on creation for the created_at field.
 	usagelog.DefaultCreatedAt = usagelogDescCreatedAt.Default.(func() time.Time)
 	userMixin := schema.User{}.Mixin()

backend/ent/schema/usage_log.go

@@ -119,6 +119,10 @@ func (UsageLog) Fields() []ent.Field {
 			Optional().
 			Nillable(),
 
+		// Cache TTL Override 标记（管理员强制替换了缓存 TTL 计费）
+		field.Bool("cache_ttl_overridden").
+			Default(false),
+
 		// 时间戳（只有 created_at，日志不可修改）
 		field.Time("created_at").
 			Default(time.Now).

backend/ent/usagelog.go

@@ -80,6 +80,8 @@ type UsageLog struct {
 	ImageCount int `json:"image_count,omitempty"`
 	// ImageSize holds the value of the "image_size" field.
 	ImageSize *string `json:"image_size,omitempty"`
+	// CacheTTLOverridden holds the value of the "cache_ttl_overridden" field.
+	CacheTTLOverridden bool `json:"cache_ttl_overridden,omitempty"`
 	// CreatedAt holds the value of the "created_at" field.
 	CreatedAt time.Time `json:"created_at,omitempty"`
 	// Edges holds the relations/edges for other nodes in the graph.
@@ -165,7 +167,7 @@ func (*UsageLog) scanValues(columns []string) ([]any, error) {
 	values := make([]any, len(columns))
 	for i := range columns {
 		switch columns[i] {
-		case usagelog.FieldStream:
+		case usagelog.FieldStream, usagelog.FieldCacheTTLOverridden:
 			values[i] = new(sql.NullBool)
 		case usagelog.FieldInputCost, usagelog.FieldOutputCost, usagelog.FieldCacheCreationCost, usagelog.FieldCacheReadCost, usagelog.FieldTotalCost, usagelog.FieldActualCost, usagelog.FieldRateMultiplier, usagelog.FieldAccountRateMultiplier:
 			values[i] = new(sql.NullFloat64)
@@ -378,6 +380,12 @@ func (_m *UsageLog) assignValues(columns []string, values []any) error {
 				_m.ImageSize = new(string)
 				*_m.ImageSize = value.String
 			}
+		case usagelog.FieldCacheTTLOverridden:
+			if value, ok := values[i].(*sql.NullBool); !ok {
+				return fmt.Errorf("unexpected type %T for field cache_ttl_overridden", values[i])
+			} else if value.Valid {
+				_m.CacheTTLOverridden = value.Bool
+			}
 		case usagelog.FieldCreatedAt:
 			if value, ok := values[i].(*sql.NullTime); !ok {
 				return fmt.Errorf("unexpected type %T for field created_at", values[i])
@@ -548,6 +556,9 @@ func (_m *UsageLog) String() string {
 		builder.WriteString(*v)
 	}
 	builder.WriteString(", ")
+	builder.WriteString("cache_ttl_overridden=")
+	builder.WriteString(fmt.Sprintf("%v", _m.CacheTTLOverridden))
+	builder.WriteString(", ")
 	builder.WriteString("created_at=")
 	builder.WriteString(_m.CreatedAt.Format(time.ANSIC))
 	builder.WriteByte(')')

backend/ent/usagelog/usagelog.go

@@ -72,6 +72,8 @@ const (
 	FieldImageCount = "image_count"
 	// FieldImageSize holds the string denoting the image_size field in the database.
 	FieldImageSize = "image_size"
+	// FieldCacheTTLOverridden holds the string denoting the cache_ttl_overridden field in the database.
+	FieldCacheTTLOverridden = "cache_ttl_overridden"
 	// FieldCreatedAt holds the string denoting the created_at field in the database.
 	FieldCreatedAt = "created_at"
 	// EdgeUser holds the string denoting the user edge name in mutations.
@@ -155,6 +157,7 @@ var Columns = []string{
 	FieldIPAddress,
 	FieldImageCount,
 	FieldImageSize,
+	FieldCacheTTLOverridden,
 	FieldCreatedAt,
 }
 
@@ -211,6 +214,8 @@ var (
 	DefaultImageCount int
 	// ImageSizeValidator is a validator for the "image_size" field. It is called by the builders before save.
 	ImageSizeValidator func(string) error
+	// DefaultCacheTTLOverridden holds the default value on creation for the "cache_ttl_overridden" field.
+	DefaultCacheTTLOverridden bool
 	// DefaultCreatedAt holds the default value on creation for the "created_at" field.
 	DefaultCreatedAt func() time.Time
 )
@@ -368,6 +373,11 @@ func ByImageSize(opts ...sql.OrderTermOption) OrderOption {
 	return sql.OrderByField(FieldImageSize, opts...).ToFunc()
 }
 
+// ByCacheTTLOverridden orders the results by the cache_ttl_overridden field.
+func ByCacheTTLOverridden(opts ...sql.OrderTermOption) OrderOption {
+	return sql.OrderByField(FieldCacheTTLOverridden, opts...).ToFunc()
+}
+
 // ByCreatedAt orders the results by the created_at field.
 func ByCreatedAt(opts ...sql.OrderTermOption) OrderOption {
 	return sql.OrderByField(FieldCreatedAt, opts...).ToFunc()

backend/ent/usagelog/where.go

@@ -200,6 +200,11 @@ func ImageSize(v string) predicate.UsageLog {
 	return predicate.UsageLog(sql.FieldEQ(FieldImageSize, v))
 }
 
+// CacheTTLOverridden applies equality check predicate on the "cache_ttl_overridden" field. It's identical to CacheTTLOverriddenEQ.
+func CacheTTLOverridden(v bool) predicate.UsageLog {
+	return predicate.UsageLog(sql.FieldEQ(FieldCacheTTLOverridden, v))
+}
+
 // CreatedAt applies equality check predicate on the "created_at" field. It's identical to CreatedAtEQ.
 func CreatedAt(v time.Time) predicate.UsageLog {
 	return predicate.UsageLog(sql.FieldEQ(FieldCreatedAt, v))
@@ -1440,6 +1445,16 @@ func ImageSizeContainsFold(v string) predicate.UsageLog {
 	return predicate.UsageLog(sql.FieldContainsFold(FieldImageSize, v))
 }
 
+// CacheTTLOverriddenEQ applies the EQ predicate on the "cache_ttl_overridden" field.
+func CacheTTLOverriddenEQ(v bool) predicate.UsageLog {
+	return predicate.UsageLog(sql.FieldEQ(FieldCacheTTLOverridden, v))
+}
+
+// CacheTTLOverriddenNEQ applies the NEQ predicate on the "cache_ttl_overridden" field.
+func CacheTTLOverriddenNEQ(v bool) predicate.UsageLog {
+	return predicate.UsageLog(sql.FieldNEQ(FieldCacheTTLOverridden, v))
+}
+
 // CreatedAtEQ applies the EQ predicate on the "created_at" field.
 func CreatedAtEQ(v time.Time) predicate.UsageLog {
 	return predicate.UsageLog(sql.FieldEQ(FieldCreatedAt, v))

backend/ent/usagelog_create.go

@@ -393,6 +393,20 @@ func (_c *UsageLogCreate) SetNillableImageSize(v *string) *UsageLogCreate {
 	return _c
 }
 
+// SetCacheTTLOverridden sets the "cache_ttl_overridden" field.
+func (_c *UsageLogCreate) SetCacheTTLOverridden(v bool) *UsageLogCreate {
+	_c.mutation.SetCacheTTLOverridden(v)
+	return _c
+}
+
+// SetNillableCacheTTLOverridden sets the "cache_ttl_overridden" field if the given value is not nil.
+func (_c *UsageLogCreate) SetNillableCacheTTLOverridden(v *bool) *UsageLogCreate {
+	if v != nil {
+		_c.SetCacheTTLOverridden(*v)
+	}
+	return _c
+}
+
 // SetCreatedAt sets the "created_at" field.
 func (_c *UsageLogCreate) SetCreatedAt(v time.Time) *UsageLogCreate {
 	_c.mutation.SetCreatedAt(v)
@@ -531,6 +545,10 @@ func (_c *UsageLogCreate) defaults() {
 		v := usagelog.DefaultImageCount
 		_c.mutation.SetImageCount(v)
 	}
+	if _, ok := _c.mutation.CacheTTLOverridden(); !ok {
+		v := usagelog.DefaultCacheTTLOverridden
+		_c.mutation.SetCacheTTLOverridden(v)
+	}
 	if _, ok := _c.mutation.CreatedAt(); !ok {
 		v := usagelog.DefaultCreatedAt()
 		_c.mutation.SetCreatedAt(v)
@@ -627,6 +645,9 @@ func (_c *UsageLogCreate) check() error {
 			return &ValidationError{Name: "image_size", err: fmt.Errorf(`ent: validator failed for field "UsageLog.image_size": %w`, err)}
 		}
 	}
+	if _, ok := _c.mutation.CacheTTLOverridden(); !ok {
+		return &ValidationError{Name: "cache_ttl_overridden", err: errors.New(`ent: missing required field "UsageLog.cache_ttl_overridden"`)}
+	}
 	if _, ok := _c.mutation.CreatedAt(); !ok {
 		return &ValidationError{Name: "created_at", err: errors.New(`ent: missing required field "UsageLog.created_at"`)}
 	}
@@ -762,6 +783,10 @@ func (_c *UsageLogCreate) createSpec() (*UsageLog, *sqlgraph.CreateSpec) {
 		_spec.SetField(usagelog.FieldImageSize, field.TypeString, value)
 		_node.ImageSize = &value
 	}
+	if value, ok := _c.mutation.CacheTTLOverridden(); ok {
+		_spec.SetField(usagelog.FieldCacheTTLOverridden, field.TypeBool, value)
+		_node.CacheTTLOverridden = value
+	}
 	if value, ok := _c.mutation.CreatedAt(); ok {
 		_spec.SetField(usagelog.FieldCreatedAt, field.TypeTime, value)
 		_node.CreatedAt = value
@@ -1407,6 +1432,18 @@ func (u *UsageLogUpsert) ClearImageSize() *UsageLogUpsert {
 	return u
 }
 
+// SetCacheTTLOverridden sets the "cache_ttl_overridden" field.
+func (u *UsageLogUpsert) SetCacheTTLOverridden(v bool) *UsageLogUpsert {
+	u.Set(usagelog.FieldCacheTTLOverridden, v)
+	return u
+}
+
+// UpdateCacheTTLOverridden sets the "cache_ttl_overridden" field to the value that was provided on create.
+func (u *UsageLogUpsert) UpdateCacheTTLOverridden() *UsageLogUpsert {
+	u.SetExcluded(usagelog.FieldCacheTTLOverridden)
+	return u
+}
+
 // UpdateNewValues updates the mutable fields using the new values that were set on create.
 // Using this option is equivalent to using:
 //
@@ -2040,6 +2077,20 @@ func (u *UsageLogUpsertOne) ClearImageSize() *UsageLogUpsertOne {
 	})
 }
 
+// SetCacheTTLOverridden sets the "cache_ttl_overridden" field.
+func (u *UsageLogUpsertOne) SetCacheTTLOverridden(v bool) *UsageLogUpsertOne {
+	return u.Update(func(s *UsageLogUpsert) {
+		s.SetCacheTTLOverridden(v)
+	})
+}
+
+// UpdateCacheTTLOverridden sets the "cache_ttl_overridden" field to the value that was provided on create.
+func (u *UsageLogUpsertOne) UpdateCacheTTLOverridden() *UsageLogUpsertOne {
+	return u.Update(func(s *UsageLogUpsert) {
+		s.UpdateCacheTTLOverridden()
+	})
+}
+
 // Exec executes the query.
 func (u *UsageLogUpsertOne) Exec(ctx context.Context) error {
 	if len(u.create.conflict) == 0 {
@@ -2839,6 +2890,20 @@ func (u *UsageLogUpsertBulk) ClearImageSize() *UsageLogUpsertBulk {
 	})
 }
 
+// SetCacheTTLOverridden sets the "cache_ttl_overridden" field.
+func (u *UsageLogUpsertBulk) SetCacheTTLOverridden(v bool) *UsageLogUpsertBulk {
+	return u.Update(func(s *UsageLogUpsert) {
+		s.SetCacheTTLOverridden(v)
+	})
+}
+
+// UpdateCacheTTLOverridden sets the "cache_ttl_overridden" field to the value that was provided on create.
+func (u *UsageLogUpsertBulk) UpdateCacheTTLOverridden() *UsageLogUpsertBulk {
+	return u.Update(func(s *UsageLogUpsert) {
+		s.UpdateCacheTTLOverridden()
+	})
+}
+
 // Exec executes the query.
 func (u *UsageLogUpsertBulk) Exec(ctx context.Context) error {
 	if u.create.err != nil {

backend/ent/usagelog_update.go

@@ -612,6 +612,20 @@ func (_u *UsageLogUpdate) ClearImageSize() *UsageLogUpdate {
 	return _u
 }
 
+// SetCacheTTLOverridden sets the "cache_ttl_overridden" field.
+func (_u *UsageLogUpdate) SetCacheTTLOverridden(v bool) *UsageLogUpdate {
+	_u.mutation.SetCacheTTLOverridden(v)
+	return _u
+}
+
+// SetNillableCacheTTLOverridden sets the "cache_ttl_overridden" field if the given value is not nil.
+func (_u *UsageLogUpdate) SetNillableCacheTTLOverridden(v *bool) *UsageLogUpdate {
+	if v != nil {
+		_u.SetCacheTTLOverridden(*v)
+	}
+	return _u
+}
+
 // SetUser sets the "user" edge to the User entity.
 func (_u *UsageLogUpdate) SetUser(v *User) *UsageLogUpdate {
 	return _u.SetUserID(v.ID)
@@ -894,6 +908,9 @@ func (_u *UsageLogUpdate) sqlSave(ctx context.Context) (_node int, err error) {
 	if _u.mutation.ImageSizeCleared() {
 		_spec.ClearField(usagelog.FieldImageSize, field.TypeString)
 	}
+	if value, ok := _u.mutation.CacheTTLOverridden(); ok {
+		_spec.SetField(usagelog.FieldCacheTTLOverridden, field.TypeBool, value)
+	}
 	if _u.mutation.UserCleared() {
 		edge := &sqlgraph.EdgeSpec{
 			Rel:     sqlgraph.M2O,
@@ -1639,6 +1656,20 @@ func (_u *UsageLogUpdateOne) ClearImageSize() *UsageLogUpdateOne {
 	return _u
 }
 
+// SetCacheTTLOverridden sets the "cache_ttl_overridden" field.
+func (_u *UsageLogUpdateOne) SetCacheTTLOverridden(v bool) *UsageLogUpdateOne {
+	_u.mutation.SetCacheTTLOverridden(v)
+	return _u
+}
+
+// SetNillableCacheTTLOverridden sets the "cache_ttl_overridden" field if the given value is not nil.
+func (_u *UsageLogUpdateOne) SetNillableCacheTTLOverridden(v *bool) *UsageLogUpdateOne {
+	if v != nil {
+		_u.SetCacheTTLOverridden(*v)
+	}
+	return _u
+}
+
 // SetUser sets the "user" edge to the User entity.
 func (_u *UsageLogUpdateOne) SetUser(v *User) *UsageLogUpdateOne {
 	return _u.SetUserID(v.ID)
@@ -1951,6 +1982,9 @@ func (_u *UsageLogUpdateOne) sqlSave(ctx context.Context) (_node *UsageLog, err
 	if _u.mutation.ImageSizeCleared() {
 		_spec.ClearField(usagelog.FieldImageSize, field.TypeString)
 	}
+	if value, ok := _u.mutation.CacheTTLOverridden(); ok {
+		_spec.SetField(usagelog.FieldCacheTTLOverridden, field.TypeBool, value)
+	}
 	if _u.mutation.UserCleared() {
 		edge := &sqlgraph.EdgeSpec{
 			Rel:     sqlgraph.M2O,

backend/internal/handler/admin/account_data.go

@@ -341,7 +341,7 @@ func (h *AccountHandler) listAccountsFiltered(ctx context.Context, platform, acc
 	pageSize := dataPageCap
 	var out []service.Account
 	for {
-		items, total, err := h.adminService.ListAccounts(ctx, page, pageSize, platform, accountType, status, search)
+		items, total, err := h.adminService.ListAccounts(ctx, page, pageSize, platform, accountType, status, search, 0)
 		if err != nil {
 			return nil, err
 		}

backend/internal/handler/admin/account_handler.go

@@ -156,7 +156,12 @@ func (h *AccountHandler) List(c *gin.Context) {
 		search = search[:100]
 	}
 
-	accounts, total, err := h.adminService.ListAccounts(c.Request.Context(), page, pageSize, platform, accountType, status, search)
+	var groupID int64
+	if groupIDStr := c.Query("group"); groupIDStr != "" {
+		groupID, _ = strconv.ParseInt(groupIDStr, 10, 64)
+	}
+
+	accounts, total, err := h.adminService.ListAccounts(c.Request.Context(), page, pageSize, platform, accountType, status, search, groupID)
 	if err != nil {
 		response.ErrorFrom(c, err)
 		return
@@ -1429,7 +1434,7 @@ func (h *AccountHandler) BatchRefreshTier(c *gin.Context) {
 	accounts := make([]*service.Account, 0)
 
 	if len(req.AccountIDs) == 0 {
-		allAccounts, _, err := h.adminService.ListAccounts(ctx, 1, 10000, "gemini", "oauth", "", "")
+		allAccounts, _, err := h.adminService.ListAccounts(ctx, 1, 10000, "gemini", "oauth", "", "", 0)
 		if err != nil {
 			response.ErrorFrom(c, err)
 			return

backend/internal/handler/admin/admin_service_stub_test.go

@@ -166,7 +166,7 @@ func (s *stubAdminService) GetGroupAPIKeys(ctx context.Context, groupID int64, p
 	return s.apiKeys, int64(len(s.apiKeys)), nil
 }
 
-func (s *stubAdminService) ListAccounts(ctx context.Context, page, pageSize int, platform, accountType, status, search string) ([]service.Account, int64, error) {
+func (s *stubAdminService) ListAccounts(ctx context.Context, page, pageSize int, platform, accountType, status, search string, groupID int64) ([]service.Account, int64, error) {
 	return s.accounts, int64(len(s.accounts)), nil
 }
 

backend/internal/handler/admin/antigravity_oauth_handler.go

@@ -65,3 +65,27 @@ func (h *AntigravityOAuthHandler) ExchangeCode(c *gin.Context) {
 
 	response.Success(c, tokenInfo)
 }
+
+// AntigravityRefreshTokenRequest represents the request for validating Antigravity refresh token
+type AntigravityRefreshTokenRequest struct {
+	RefreshToken string `json:"refresh_token" binding:"required"`
+	ProxyID      *int64 `json:"proxy_id"`
+}
+
+// RefreshToken validates an Antigravity refresh token and returns full token info
+// POST /api/v1/admin/antigravity/oauth/refresh-token
+func (h *AntigravityOAuthHandler) RefreshToken(c *gin.Context) {
+	var req AntigravityRefreshTokenRequest
+	if err := c.ShouldBindJSON(&req); err != nil {
+		response.BadRequest(c, "请求无效: "+err.Error())
+		return
+	}
+
+	tokenInfo, err := h.antigravityOAuthService.ValidateRefreshToken(c.Request.Context(), req.RefreshToken, req.ProxyID)
+	if err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
+
+	response.Success(c, tokenInfo)
+}

backend/internal/handler/admin/redeem_handler.go

@@ -202,7 +202,7 @@ func (h *RedeemHandler) Export(c *gin.Context) {
 	writer := csv.NewWriter(&buf)
 
 	// Write header
-	if err := writer.Write([]string{"id", "code", "type", "value", "status", "used_by", "used_at", "created_at"}); err != nil {
+	if err := writer.Write([]string{"id", "code", "type", "value", "status", "used_by", "used_by_email", "used_at", "created_at"}); err != nil {
 		response.InternalError(c, "Failed to export redeem codes: "+err.Error())
 		return
 	}
@@ -213,6 +213,10 @@ func (h *RedeemHandler) Export(c *gin.Context) {
 		if code.UsedBy != nil {
 			usedBy = fmt.Sprintf("%d", *code.UsedBy)
 		}
+		usedByEmail := ""
+		if code.User != nil {
+			usedByEmail = code.User.Email
+		}
 		usedAt := ""
 		if code.UsedAt != nil {
 			usedAt = code.UsedAt.Format("2006-01-02 15:04:05")
@@ -224,6 +228,7 @@ func (h *RedeemHandler) Export(c *gin.Context) {
 			fmt.Sprintf("%.2f", code.Value),
 			code.Status,
 			usedBy,
+			usedByEmail,
 			usedAt,
 			code.CreatedAt.Format("2006-01-02 15:04:05"),
 		}); err != nil {

backend/internal/handler/dto/mappers.go

@@ -211,6 +211,13 @@ func AccountFromServiceShallow(a *service.Account) *Account {
 			enabled := true
 			out.EnableSessionIDMasking = &enabled
 		}
+		// 缓存 TTL 强制替换
+		if a.IsCacheTTLOverrideEnabled() {
+			enabled := true
+			out.CacheTTLOverrideEnabled = &enabled
+			target := a.GetCacheTTLOverrideTarget()
+			out.CacheTTLOverrideTarget = &target
+		}
 	}
 
 	return out
@@ -398,6 +405,7 @@ func usageLogFromServiceUser(l *service.UsageLog) UsageLog {
 		ImageCount:            l.ImageCount,
 		ImageSize:             l.ImageSize,
 		UserAgent:             l.UserAgent,
+		CacheTTLOverridden:    l.CacheTTLOverridden,
 		CreatedAt:             l.CreatedAt,
 		User:                  UserFromServiceShallow(l.User),
 		APIKey:                APIKeyFromService(l.APIKey),

backend/internal/handler/dto/types.go

@@ -150,6 +150,11 @@ type Account struct {
 	// 从 extra 字段提取，方便前端显示和编辑
 	EnableSessionIDMasking *bool `json:"session_id_masking_enabled,omitempty"`
 
+	// 缓存 TTL 强制替换（仅 Anthropic OAuth/SetupToken 账号有效）
+	// 启用后将所有 cache creation tokens 归入指定的 TTL 类型计费
+	CacheTTLOverrideEnabled *bool   `json:"cache_ttl_override_enabled,omitempty"`
+	CacheTTLOverrideTarget  *string `json:"cache_ttl_override_target,omitempty"`
+
 	Proxy         *Proxy         `json:"proxy,omitempty"`
 	AccountGroups []AccountGroup `json:"account_groups,omitempty"`
 
@@ -273,6 +278,9 @@ type UsageLog struct {
 	// User-Agent
 	UserAgent *string `json:"user_agent"`
 
+	// Cache TTL Override 标记
+	CacheTTLOverridden bool `json:"cache_ttl_overridden"`
+
 	CreatedAt time.Time `json:"created_at"`
 
 	User         *User             `json:"user,omitempty"`

backend/internal/pkg/antigravity/claude_types.go

@@ -27,7 +27,7 @@ type ClaudeMessage struct {
 
 // ThinkingConfig Thinking 配置
 type ThinkingConfig struct {
-	Type         string `json:"type"`                    // "enabled" or "disabled"
+	Type         string `json:"type"`                    // "enabled" / "adaptive" / "disabled"
 	BudgetTokens int    `json:"budget_tokens,omitempty"` // thinking budget
 }
 

backend/internal/pkg/antigravity/gemini_types.go

@@ -155,6 +155,7 @@ type GeminiUsageMetadata struct {
 	CandidatesTokenCount    int `json:"candidatesTokenCount,omitempty"`
 	CachedContentTokenCount int `json:"cachedContentTokenCount,omitempty"`
 	TotalTokenCount         int `json:"totalTokenCount,omitempty"`
+	ThoughtsTokenCount      int `json:"thoughtsTokenCount,omitempty"` // thinking tokens（按输出价格计费）
 }
 
 // GeminiGroundingMetadata Gemini grounding 元数据（Web Search）

backend/internal/pkg/antigravity/request_transformer.go

@@ -64,6 +64,10 @@ const MaxTokensBudgetPadding = 1000
 // Gemini 2.5 Flash thinking budget 上限
 const Gemini25FlashThinkingBudgetLimit = 24576
 
+// 对于 Antigravity 的 Claude（budget-only）模型，该语义最终等价为 thinkingBudget=24576。
+// 这里复用相同数值以保持行为一致。
+const ClaudeAdaptiveHighThinkingBudgetTokens = Gemini25FlashThinkingBudgetLimit
+
 // ensureMaxTokensGreaterThanBudget 确保 max_tokens > budget_tokens
 // Claude API 要求启用 thinking 时，max_tokens 必须大于 thinking.budget_tokens
 // 返回调整后的 maxTokens 和是否进行了调整
@@ -96,7 +100,7 @@ func TransformClaudeToGeminiWithOptions(claudeReq *ClaudeRequest, projectID, map
 	}
 
 	// 检测是否启用 thinking
-	isThinkingEnabled := claudeReq.Thinking != nil && claudeReq.Thinking.Type == "enabled"
+	isThinkingEnabled := claudeReq.Thinking != nil && (claudeReq.Thinking.Type == "enabled" || claudeReq.Thinking.Type == "adaptive")
 
 	// 只有 Gemini 模型支持 dummy thought workaround
 	// Claude 模型通过 Vertex/Google API 需要有效的 thought signatures
@@ -198,8 +202,7 @@ type modelInfo struct {
 
 // modelInfoMap 模型前缀 → 模型信息映射
 // 只有在此映射表中的模型才会注入身份提示词
-// 注意：当前 claude-opus-4-6 会被映射到 claude-opus-4-5-thinking，
+// 注意：模型映射逻辑在网关层完成；这里仅用于按模型前缀判断是否注入身份提示词。
-// 但保留此条目以便后续 Antigravity 上游支持 4.6 时快速切换
 var modelInfoMap = map[string]modelInfo{
 	"claude-opus-4-5":   {DisplayName: "Claude Opus 4.5", CanonicalID: "claude-opus-4-5-20250929"},
 	"claude-opus-4-6":   {DisplayName: "Claude Opus 4.6", CanonicalID: "claude-opus-4-6"},
@@ -593,6 +596,10 @@ func maxOutputTokensLimit(model string) int {
 	return maxOutputTokensUpperBound
 }
 
+func isAntigravityOpus46Model(model string) bool {
+	return strings.HasPrefix(strings.ToLower(model), "claude-opus-4-6")
+}
+
 func buildGenerationConfig(req *ClaudeRequest) *GeminiGenerationConfig {
 	maxLimit := maxOutputTokensLimit(req.Model)
 	config := &GeminiGenerationConfig{
@@ -606,25 +613,36 @@ func buildGenerationConfig(req *ClaudeRequest) *GeminiGenerationConfig {
 	}
 
 	// Thinking 配置
-	if req.Thinking != nil && req.Thinking.Type == "enabled" {
+	if req.Thinking != nil && (req.Thinking.Type == "enabled" || req.Thinking.Type == "adaptive") {
 		config.ThinkingConfig = &GeminiThinkingConfig{
 			IncludeThoughts: true,
 		}
+
+		// - thinking.type=enabled：budget_tokens>0 用显式预算
+		// - thinking.type=adaptive：仅在 Antigravity 的 Opus 4.6 上覆写为 （24576）
+		budget := -1
 		if req.Thinking.BudgetTokens > 0 {
-			budget := req.Thinking.BudgetTokens
+			budget = req.Thinking.BudgetTokens
+		}
+		if req.Thinking.Type == "adaptive" && isAntigravityOpus46Model(req.Model) {
+			budget = ClaudeAdaptiveHighThinkingBudgetTokens
+		}
+
+		// 正预算需要做上限与 max_tokens 约束；动态预算（-1）直接透传给上游。
+		if budget > 0 {
 			// gemini-2.5-flash 上限
 			if strings.Contains(req.Model, "gemini-2.5-flash") && budget > Gemini25FlashThinkingBudgetLimit {
 				budget = Gemini25FlashThinkingBudgetLimit
 			}
-			config.ThinkingConfig.ThinkingBudget = budget
 
-			// 自动修正：max_tokens 必须大于 budget_tokens
+			// 自动修正：max_tokens 必须大于 budget_tokens（Claude 上游要求）
 			if adjusted, ok := ensureMaxTokensGreaterThanBudget(config.MaxOutputTokens, budget); ok {
 				log.Printf("[Antigravity] Auto-adjusted max_tokens from %d to %d (must be > budget_tokens=%d)",
 					config.MaxOutputTokens, adjusted, budget)
 				config.MaxOutputTokens = adjusted
 			}
 		}
+		config.ThinkingConfig.ThinkingBudget = budget
 	}
 
 	if config.MaxOutputTokens > maxLimit {

backend/internal/pkg/antigravity/request_transformer_test.go

@@ -259,3 +259,93 @@ func TestBuildTools_CustomTypeTools(t *testing.T) {
 		})
 	}
 }
+
+func TestBuildGenerationConfig_ThinkingDynamicBudget(t *testing.T) {
+	tests := []struct {
+		name        string
+		model       string
+		thinking    *ThinkingConfig
+		wantBudget  int
+		wantPresent bool
+	}{
+		{
+			name:        "enabled without budget defaults to dynamic (-1)",
+			model:       "claude-opus-4-6-thinking",
+			thinking:    &ThinkingConfig{Type: "enabled"},
+			wantBudget:  -1,
+			wantPresent: true,
+		},
+		{
+			name:        "enabled with budget uses the provided value",
+			model:       "claude-opus-4-6-thinking",
+			thinking:    &ThinkingConfig{Type: "enabled", BudgetTokens: 1024},
+			wantBudget:  1024,
+			wantPresent: true,
+		},
+		{
+			name:        "enabled with -1 budget uses dynamic (-1)",
+			model:       "claude-opus-4-6-thinking",
+			thinking:    &ThinkingConfig{Type: "enabled", BudgetTokens: -1},
+			wantBudget:  -1,
+			wantPresent: true,
+		},
+		{
+			name:        "adaptive on opus4.6 maps to high budget (24576)",
+			model:       "claude-opus-4-6-thinking",
+			thinking:    &ThinkingConfig{Type: "adaptive", BudgetTokens: 20000},
+			wantBudget:  ClaudeAdaptiveHighThinkingBudgetTokens,
+			wantPresent: true,
+		},
+		{
+			name:        "adaptive on non-opus model keeps default dynamic (-1)",
+			model:       "claude-sonnet-4-5-thinking",
+			thinking:    &ThinkingConfig{Type: "adaptive"},
+			wantBudget:  -1,
+			wantPresent: true,
+		},
+		{
+			name:        "disabled does not emit thinkingConfig",
+			model:       "claude-opus-4-6-thinking",
+			thinking:    &ThinkingConfig{Type: "disabled", BudgetTokens: 1024},
+			wantBudget:  0,
+			wantPresent: false,
+		},
+		{
+			name:        "nil thinking does not emit thinkingConfig",
+			model:       "claude-opus-4-6-thinking",
+			thinking:    nil,
+			wantBudget:  0,
+			wantPresent: false,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			req := &ClaudeRequest{
+				Model:    tt.model,
+				Thinking: tt.thinking,
+			}
+			cfg := buildGenerationConfig(req)
+			if cfg == nil {
+				t.Fatalf("expected non-nil generationConfig")
+			}
+
+			if tt.wantPresent {
+				if cfg.ThinkingConfig == nil {
+					t.Fatalf("expected thinkingConfig to be present")
+				}
+				if !cfg.ThinkingConfig.IncludeThoughts {
+					t.Fatalf("expected includeThoughts=true")
+				}
+				if cfg.ThinkingConfig.ThinkingBudget != tt.wantBudget {
+					t.Fatalf("expected thinkingBudget=%d, got %d", tt.wantBudget, cfg.ThinkingConfig.ThinkingBudget)
+				}
+				return
+			}
+
+			if cfg.ThinkingConfig != nil {
+				t.Fatalf("expected thinkingConfig to be nil, got %+v", cfg.ThinkingConfig)
+			}
+		})
+	}
+}

backend/internal/pkg/antigravity/response_transformer.go

@@ -279,7 +279,7 @@ func (p *NonStreamingProcessor) buildResponse(geminiResp *GeminiResponse, respon
 	if geminiResp.UsageMetadata != nil {
 		cached := geminiResp.UsageMetadata.CachedContentTokenCount
 		usage.InputTokens = geminiResp.UsageMetadata.PromptTokenCount - cached
-		usage.OutputTokens = geminiResp.UsageMetadata.CandidatesTokenCount
+		usage.OutputTokens = geminiResp.UsageMetadata.CandidatesTokenCount + geminiResp.UsageMetadata.ThoughtsTokenCount
 		usage.CacheReadInputTokens = cached
 	}
 

backend/internal/pkg/antigravity/stream_transformer.go

@@ -85,7 +85,7 @@ func (p *StreamingProcessor) ProcessLine(line string) []byte {
 	if geminiResp.UsageMetadata != nil {
 		cached := geminiResp.UsageMetadata.CachedContentTokenCount
 		p.inputTokens = geminiResp.UsageMetadata.PromptTokenCount - cached
-		p.outputTokens = geminiResp.UsageMetadata.CandidatesTokenCount
+		p.outputTokens = geminiResp.UsageMetadata.CandidatesTokenCount + geminiResp.UsageMetadata.ThoughtsTokenCount
 		p.cacheReadTokens = cached
 	}
 
@@ -146,7 +146,7 @@ func (p *StreamingProcessor) emitMessageStart(v1Resp *V1InternalResponse) []byte
 	if v1Resp.Response.UsageMetadata != nil {
 		cached := v1Resp.Response.UsageMetadata.CachedContentTokenCount
 		usage.InputTokens = v1Resp.Response.UsageMetadata.PromptTokenCount - cached
-		usage.OutputTokens = v1Resp.Response.UsageMetadata.CandidatesTokenCount
+		usage.OutputTokens = v1Resp.Response.UsageMetadata.CandidatesTokenCount + v1Resp.Response.UsageMetadata.ThoughtsTokenCount
 		usage.CacheReadInputTokens = cached
 	}
 

backend/internal/pkg/claude/constants.go

@@ -10,6 +10,7 @@ const (
 	BetaInterleavedThinking      = "interleaved-thinking-2025-05-14"
 	BetaFineGrainedToolStreaming = "fine-grained-tool-streaming-2025-05-14"
 	BetaTokenCounting            = "token-counting-2024-11-01"
+	BetaContext1M                = "context-1m-2025-08-07"
 )
 
 // DefaultBetaHeader Claude Code 客户端默认的 anthropic-beta header
@@ -77,6 +78,12 @@ var DefaultModels = []Model{
 		DisplayName: "Claude Opus 4.6",
 		CreatedAt:   "2026-02-06T00:00:00Z",
 	},
+	{
+		ID:          "claude-sonnet-4-6",
+		Type:        "model",
+		DisplayName: "Claude Sonnet 4.6",
+		CreatedAt:   "2026-02-18T00:00:00Z",
+	},
 	{
 		ID:          "claude-sonnet-4-5-20250929",
 		Type:        "model",

backend/internal/repository/account_repo.go

@@ -435,10 +435,10 @@ func (r *accountRepository) Delete(ctx context.Context, id int64) error {
 }
 
 func (r *accountRepository) List(ctx context.Context, params pagination.PaginationParams) ([]service.Account, *pagination.PaginationResult, error) {
-	return r.ListWithFilters(ctx, params, "", "", "", "")
+	return r.ListWithFilters(ctx, params, "", "", "", "", 0)
 }
 
-func (r *accountRepository) ListWithFilters(ctx context.Context, params pagination.PaginationParams, platform, accountType, status, search string) ([]service.Account, *pagination.PaginationResult, error) {
+func (r *accountRepository) ListWithFilters(ctx context.Context, params pagination.PaginationParams, platform, accountType, status, search string, groupID int64) ([]service.Account, *pagination.PaginationResult, error) {
 	q := r.client.Account.Query()
 
 	if platform != "" {
@@ -448,11 +448,19 @@ func (r *accountRepository) ListWithFilters(ctx context.Context, params paginati
 		q = q.Where(dbaccount.TypeEQ(accountType))
 	}
 	if status != "" {
-		q = q.Where(dbaccount.StatusEQ(status))
+		switch status {
+		case "rate_limited":
+			q = q.Where(dbaccount.RateLimitResetAtGT(time.Now()))
+		default:
+			q = q.Where(dbaccount.StatusEQ(status))
+		}
 	}
 	if search != "" {
 		q = q.Where(dbaccount.NameContainsFold(search))
 	}
+	if groupID > 0 {
+		q = q.Where(dbaccount.HasAccountGroupsWith(dbaccountgroup.GroupIDEQ(groupID)))
+	}
 
 	total, err := q.Count(ctx)
 	if err != nil {

backend/internal/repository/account_repo_integration_test.go

@@ -238,7 +238,7 @@ func (s *AccountRepoSuite) TestListWithFilters() {
 
 			tt.setup(client)
 
-			accounts, _, err := repo.ListWithFilters(ctx, pagination.PaginationParams{Page: 1, PageSize: 10}, tt.platform, tt.accType, tt.status, tt.search)
+			accounts, _, err := repo.ListWithFilters(ctx, pagination.PaginationParams{Page: 1, PageSize: 10}, tt.platform, tt.accType, tt.status, tt.search, 0)
 			s.Require().NoError(err)
 			s.Require().Len(accounts, tt.wantCount)
 			if tt.validate != nil {
@@ -305,7 +305,7 @@ func (s *AccountRepoSuite) TestPreload_And_VirtualFields() {
 	s.Require().Len(got.Groups, 1, "expected Groups to be populated")
 	s.Require().Equal(group.ID, got.Groups[0].ID)
 
-	accounts, page, err := s.repo.ListWithFilters(s.ctx, pagination.PaginationParams{Page: 1, PageSize: 10}, "", "", "", "acc")
+	accounts, page, err := s.repo.ListWithFilters(s.ctx, pagination.PaginationParams{Page: 1, PageSize: 10}, "", "", "", "acc", 0)
 	s.Require().NoError(err, "ListWithFilters")
 	s.Require().Equal(int64(1), page.Total)
 	s.Require().Len(accounts, 1)

backend/internal/repository/redeem_code_repo.go

@@ -6,6 +6,7 @@ import (
 
 	dbent "github.com/Wei-Shaw/sub2api/ent"
 	"github.com/Wei-Shaw/sub2api/ent/redeemcode"
+	"github.com/Wei-Shaw/sub2api/ent/user"
 	"github.com/Wei-Shaw/sub2api/internal/pkg/pagination"
 	"github.com/Wei-Shaw/sub2api/internal/service"
 )
@@ -106,7 +107,12 @@ func (r *redeemCodeRepository) ListWithFilters(ctx context.Context, params pagin
 		q = q.Where(redeemcode.StatusEQ(status))
 	}
 	if search != "" {
-		q = q.Where(redeemcode.CodeContainsFold(search))
+		q = q.Where(
+			redeemcode.Or(
+				redeemcode.CodeContainsFold(search),
+				redeemcode.HasUserWith(user.EmailContainsFold(search)),
+			),
+		)
 	}
 
 	total, err := q.Count(ctx)

backend/internal/repository/usage_log_repo.go

@@ -22,7 +22,7 @@ import (
 	"github.com/lib/pq"
 )
 
-const usageLogSelectColumns = "id, user_id, api_key_id, account_id, request_id, model, group_id, subscription_id, input_tokens, output_tokens, cache_creation_tokens, cache_read_tokens, cache_creation_5m_tokens, cache_creation_1h_tokens, input_cost, output_cost, cache_creation_cost, cache_read_cost, total_cost, actual_cost, rate_multiplier, account_rate_multiplier, billing_type, stream, duration_ms, first_token_ms, user_agent, ip_address, image_count, image_size, reasoning_effort, created_at"
+const usageLogSelectColumns = "id, user_id, api_key_id, account_id, request_id, model, group_id, subscription_id, input_tokens, output_tokens, cache_creation_tokens, cache_read_tokens, cache_creation_5m_tokens, cache_creation_1h_tokens, input_cost, output_cost, cache_creation_cost, cache_read_cost, total_cost, actual_cost, rate_multiplier, account_rate_multiplier, billing_type, stream, duration_ms, first_token_ms, user_agent, ip_address, image_count, image_size, reasoning_effort, cache_ttl_overridden, created_at"
 
 type usageLogRepository struct {
 	client *dbent.Client
@@ -115,6 +115,7 @@ func (r *usageLogRepository) Create(ctx context.Context, log *service.UsageLog)
 				image_count,
 				image_size,
 				reasoning_effort,
+				cache_ttl_overridden,
 				created_at
 			) VALUES (
 				$1, $2, $3, $4, $5,
@@ -122,7 +123,7 @@ func (r *usageLogRepository) Create(ctx context.Context, log *service.UsageLog)
 				$8, $9, $10, $11,
 				$12, $13,
 				$14, $15, $16, $17, $18, $19,
-				$20, $21, $22, $23, $24, $25, $26, $27, $28, $29, $30, $31
+				$20, $21, $22, $23, $24, $25, $26, $27, $28, $29, $30, $31, $32
 			)
 			ON CONFLICT (request_id, api_key_id) DO NOTHING
 			RETURNING id, created_at
@@ -173,6 +174,7 @@ func (r *usageLogRepository) Create(ctx context.Context, log *service.UsageLog)
 		log.ImageCount,
 		imageSize,
 		reasoningEffort,
+		log.CacheTTLOverridden,
 		createdAt,
 	}
 	if err := scanSingleRow(ctx, sqlq, query, args, &log.ID, &log.CreatedAt); err != nil {
@@ -2195,6 +2197,7 @@ func scanUsageLog(scanner interface{ Scan(...any) error }) (*service.UsageLog, e
 		imageCount            int
 		imageSize             sql.NullString
 		reasoningEffort       sql.NullString
+		cacheTTLOverridden    bool
 		createdAt             time.Time
 	)
 
@@ -2230,6 +2233,7 @@ func scanUsageLog(scanner interface{ Scan(...any) error }) (*service.UsageLog, e
 		&imageCount,
 		&imageSize,
 		&reasoningEffort,
+		&cacheTTLOverridden,
 		&createdAt,
 	); err != nil {
 		return nil, err
@@ -2258,6 +2262,7 @@ func scanUsageLog(scanner interface{ Scan(...any) error }) (*service.UsageLog, e
 		BillingType:           int8(billingType),
 		Stream:                stream,
 		ImageCount:            imageCount,
+		CacheTTLOverridden:    cacheTTLOverridden,
 		CreatedAt:             createdAt,
 	}
 

backend/internal/repository/user_repo.go

@@ -10,6 +10,7 @@ import (
 	"time"
 
 	dbent "github.com/Wei-Shaw/sub2api/ent"
+	"github.com/Wei-Shaw/sub2api/ent/apikey"
 	dbuser "github.com/Wei-Shaw/sub2api/ent/user"
 	"github.com/Wei-Shaw/sub2api/ent/userallowedgroup"
 	"github.com/Wei-Shaw/sub2api/ent/usersubscription"
@@ -191,6 +192,7 @@ func (r *userRepository) ListWithFilters(ctx context.Context, params pagination.
 				dbuser.EmailContainsFold(filters.Search),
 				dbuser.UsernameContainsFold(filters.Search),
 				dbuser.NotesContainsFold(filters.Search),
+				dbuser.HasAPIKeysWith(apikey.KeyContainsFold(filters.Search)),
 			),
 		)
 	}

backend/internal/server/api_contract_test.go

@@ -401,6 +401,7 @@ func TestAPIContracts(t *testing.T) {
 							"first_token_ms": 50,
 							"image_count": 0,
 							"image_size": null,
+							"cache_ttl_overridden": false,
 							"created_at": "2025-01-02T03:04:05Z",
 							"user_agent": null
 						}
@@ -936,7 +937,7 @@ func (s *stubAccountRepo) List(ctx context.Context, params pagination.Pagination
 	return nil, nil, errors.New("not implemented")
 }
 
-func (s *stubAccountRepo) ListWithFilters(ctx context.Context, params pagination.PaginationParams, platform, accountType, status, search string) ([]service.Account, *pagination.PaginationResult, error) {
+func (s *stubAccountRepo) ListWithFilters(ctx context.Context, params pagination.PaginationParams, platform, accountType, status, search string, groupID int64) ([]service.Account, *pagination.PaginationResult, error) {
 	return nil, nil, errors.New("not implemented")
 }
 

backend/internal/server/middleware/cors.go

@@ -70,7 +70,15 @@ func CORS(cfg config.CORSConfig) gin.HandlerFunc {
 			}
 		}
 
-		c.Writer.Header().Set("Access-Control-Allow-Headers", "Content-Type, Content-Length, Accept-Encoding, X-CSRF-Token, Authorization, accept, origin, Cache-Control, X-Requested-With, X-API-Key")
+		allowHeaders := []string{"Content-Type", "Content-Length", "Accept-Encoding", "X-CSRF-Token", "Authorization", "accept", "origin", "Cache-Control", "X-Requested-With", "X-API-Key"}
+
+		// openai node sdk
+		openAIProperties := []string{"lang", "package-version", "os", "arch", "retry-count", "runtime", "runtime-version", "async", "helper-method", "poll-helper", "custom-poll-interval", "timeout"}
+		for _, prop := range openAIProperties {
+			allowHeaders = append(allowHeaders, "x-stainless-"+prop)
+		}
+
+		c.Writer.Header().Set("Access-Control-Allow-Headers", strings.Join(allowHeaders, ", "))
 		c.Writer.Header().Set("Access-Control-Allow-Methods", "POST, OPTIONS, GET, PUT, DELETE, PATCH")
 
 		// 处理预检请求

backend/internal/server/routes/admin.go

@@ -281,6 +281,7 @@ func registerAntigravityOAuthRoutes(admin *gin.RouterGroup, h *handler.Handlers)
 	{
 		antigravity.POST("/oauth/auth-url", h.Admin.AntigravityOAuth.GenerateAuthURL)
 		antigravity.POST("/oauth/exchange-code", h.Admin.AntigravityOAuth.ExchangeCode)
+		antigravity.POST("/oauth/refresh-token", h.Admin.AntigravityOAuth.RefreshToken)
 	}
 }
 

backend/internal/service/account.go

@@ -752,6 +752,38 @@ func (a *Account) IsSessionIDMaskingEnabled() bool {
 	return false
 }
 
+// IsCacheTTLOverrideEnabled 检查是否启用缓存 TTL 强制替换
+// 仅适用于 Anthropic OAuth/SetupToken 类型账号
+// 启用后将所有 cache creation tokens 归入指定的 TTL 类型（5m 或 1h）
+func (a *Account) IsCacheTTLOverrideEnabled() bool {
+	if !a.IsAnthropicOAuthOrSetupToken() {
+		return false
+	}
+	if a.Extra == nil {
+		return false
+	}
+	if v, ok := a.Extra["cache_ttl_override_enabled"]; ok {
+		if enabled, ok := v.(bool); ok {
+			return enabled
+		}
+	}
+	return false
+}
+
+// GetCacheTTLOverrideTarget 获取缓存 TTL 强制替换的目标类型
+// 返回 "5m" 或 "1h"，默认 "5m"
+func (a *Account) GetCacheTTLOverrideTarget() string {
+	if a.Extra == nil {
+		return "5m"
+	}
+	if v, ok := a.Extra["cache_ttl_override_target"]; ok {
+		if target, ok := v.(string); ok && (target == "5m" || target == "1h") {
+			return target
+		}
+	}
+	return "5m"
+}
+
 // GetWindowCostLimit 获取 5h 窗口费用阈值（美元）
 // 返回 0 表示未启用
 func (a *Account) GetWindowCostLimit() float64 {

backend/internal/service/account_service.go

@@ -32,7 +32,7 @@ type AccountRepository interface {
 	Delete(ctx context.Context, id int64) error
 
 	List(ctx context.Context, params pagination.PaginationParams) ([]Account, *pagination.PaginationResult, error)
-	ListWithFilters(ctx context.Context, params pagination.PaginationParams, platform, accountType, status, search string) ([]Account, *pagination.PaginationResult, error)
+	ListWithFilters(ctx context.Context, params pagination.PaginationParams, platform, accountType, status, search string, groupID int64) ([]Account, *pagination.PaginationResult, error)
 	ListByGroup(ctx context.Context, groupID int64) ([]Account, error)
 	ListActive(ctx context.Context) ([]Account, error)
 	ListByPlatform(ctx context.Context, platform string) ([]Account, error)

backend/internal/service/account_service_delete_test.go

@@ -75,7 +75,7 @@ func (s *accountRepoStub) List(ctx context.Context, params pagination.Pagination
 	panic("unexpected List call")
 }
 
-func (s *accountRepoStub) ListWithFilters(ctx context.Context, params pagination.PaginationParams, platform, accountType, status, search string) ([]Account, *pagination.PaginationResult, error) {
+func (s *accountRepoStub) ListWithFilters(ctx context.Context, params pagination.PaginationParams, platform, accountType, status, search string, groupID int64) ([]Account, *pagination.PaginationResult, error) {
 	panic("unexpected ListWithFilters call")
 }
 

backend/internal/service/admin_service.go

@@ -39,7 +39,7 @@ type AdminService interface {
 	UpdateGroupSortOrders(ctx context.Context, updates []GroupSortOrderUpdate) error
 
 	// Account management
-	ListAccounts(ctx context.Context, page, pageSize int, platform, accountType, status, search string) ([]Account, int64, error)
+	ListAccounts(ctx context.Context, page, pageSize int, platform, accountType, status, search string, groupID int64) ([]Account, int64, error)
 	GetAccount(ctx context.Context, id int64) (*Account, error)
 	GetAccountsByIDs(ctx context.Context, ids []int64) ([]*Account, error)
 	CreateAccount(ctx context.Context, input *CreateAccountInput) (*Account, error)
@@ -1021,9 +1021,9 @@ func (s *adminServiceImpl) UpdateGroupSortOrders(ctx context.Context, updates []
 }
 
 // Account management implementations
-func (s *adminServiceImpl) ListAccounts(ctx context.Context, page, pageSize int, platform, accountType, status, search string) ([]Account, int64, error) {
+func (s *adminServiceImpl) ListAccounts(ctx context.Context, page, pageSize int, platform, accountType, status, search string, groupID int64) ([]Account, int64, error) {
 	params := pagination.PaginationParams{Page: page, PageSize: pageSize}
-	accounts, result, err := s.accountRepo.ListWithFilters(ctx, params, platform, accountType, status, search)
+	accounts, result, err := s.accountRepo.ListWithFilters(ctx, params, platform, accountType, status, search, groupID)
 	if err != nil {
 		return nil, 0, err
 	}

backend/internal/service/admin_service_search_test.go

@@ -24,7 +24,7 @@ type accountRepoStubForAdminList struct {
 	listWithFiltersErr      error
 }
 
-func (s *accountRepoStubForAdminList) ListWithFilters(_ context.Context, params pagination.PaginationParams, platform, accountType, status, search string) ([]Account, *pagination.PaginationResult, error) {
+func (s *accountRepoStubForAdminList) ListWithFilters(_ context.Context, params pagination.PaginationParams, platform, accountType, status, search string, groupID int64) ([]Account, *pagination.PaginationResult, error) {
 	s.listWithFiltersCalls++
 	s.listWithFiltersParams = params
 	s.listWithFiltersPlatform = platform
@@ -168,7 +168,7 @@ func TestAdminService_ListAccounts_WithSearch(t *testing.T) {
 		}
 		svc := &adminServiceImpl{accountRepo: repo}
 
-		accounts, total, err := svc.ListAccounts(context.Background(), 1, 20, PlatformGemini, AccountTypeOAuth, StatusActive, "acc")
+		accounts, total, err := svc.ListAccounts(context.Background(), 1, 20, PlatformGemini, AccountTypeOAuth, StatusActive, "acc", 0)
 		require.NoError(t, err)
 		require.Equal(t, int64(10), total)
 		require.Equal(t, []Account{{ID: 1, Name: "acc"}}, accounts)

backend/internal/service/antigravity_gateway_service.go

@@ -1309,7 +1309,7 @@ func (s *AntigravityGatewayService) Forward(ctx context.Context, c *gin.Context,
 		return nil, s.writeClaudeError(c, http.StatusForbidden, "permission_error", fmt.Sprintf("model %s not in whitelist", claudeReq.Model))
 	}
 	// 应用 thinking 模式自动后缀：如果 thinking 开启且目标是 claude-sonnet-4-5，自动改为 thinking 版本
-	thinkingEnabled := claudeReq.Thinking != nil && claudeReq.Thinking.Type == "enabled"
+	thinkingEnabled := claudeReq.Thinking != nil && (claudeReq.Thinking.Type == "enabled" || claudeReq.Thinking.Type == "adaptive")
 	mappedModel = applyThinkingModelSuffix(mappedModel, thinkingEnabled)
 
 	// 获取 access_token
@@ -4111,6 +4111,15 @@ func (s *AntigravityGatewayService) extractSSEUsage(line string, usage *ClaudeUs
 	if v, ok := u["cache_creation_input_tokens"].(float64); ok && int(v) > 0 {
 		usage.CacheCreationInputTokens = int(v)
 	}
+	// 解析嵌套的 cache_creation 对象中的 5m/1h 明细
+	if cc, ok := u["cache_creation"].(map[string]any); ok {
+		if v, ok := cc["ephemeral_5m_input_tokens"].(float64); ok {
+			usage.CacheCreation5mTokens = int(v)
+		}
+		if v, ok := cc["ephemeral_1h_input_tokens"].(float64); ok {
+			usage.CacheCreation1hTokens = int(v)
+		}
+	}
 }
 
 // extractClaudeUsage 从非流式 Claude 响应提取 usage
@@ -4133,6 +4142,15 @@ func (s *AntigravityGatewayService) extractClaudeUsage(body []byte) *ClaudeUsage
 		if v, ok := u["cache_creation_input_tokens"].(float64); ok {
 			usage.CacheCreationInputTokens = int(v)
 		}
+		// 解析嵌套的 cache_creation 对象中的 5m/1h 明细
+		if cc, ok := u["cache_creation"].(map[string]any); ok {
+			if v, ok := cc["ephemeral_5m_input_tokens"].(float64); ok {
+				usage.CacheCreation5mTokens = int(v)
+			}
+			if v, ok := cc["ephemeral_1h_input_tokens"].(float64); ok {
+				usage.CacheCreation1hTokens = int(v)
+			}
+		}
 	}
 	return usage
 }

backend/internal/service/antigravity_gateway_service_test.go

@@ -553,6 +553,75 @@ func TestHandleClaudeStreamingResponse_NormalComplete(t *testing.T) {
 	require.NotContains(t, body, "event: error")
 }
 
+// TestHandleGeminiStreamingResponse_ThoughtsTokenCount
+// 验证：Gemini 流式转发时 thoughtsTokenCount 被计入 OutputTokens
+func TestHandleGeminiStreamingResponse_ThoughtsTokenCount(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	svc := newAntigravityTestService(&config.Config{
+		Gateway: config.GatewayConfig{MaxLineSize: defaultMaxLineSize},
+	})
+
+	rec := httptest.NewRecorder()
+	c, _ := gin.CreateTestContext(rec)
+	c.Request = httptest.NewRequest(http.MethodPost, "/", nil)
+
+	pr, pw := io.Pipe()
+	resp := &http.Response{StatusCode: http.StatusOK, Body: pr, Header: http.Header{}}
+
+	go func() {
+		defer func() { _ = pw.Close() }()
+		fmt.Fprintln(pw, `data: {"candidates":[{"content":{"parts":[{"text":"Hello"}]}}],"usageMetadata":{"promptTokenCount":100,"candidatesTokenCount":20,"thoughtsTokenCount":50}}`)
+		fmt.Fprintln(pw, "")
+		fmt.Fprintln(pw, `data: {"candidates":[{"content":{"parts":[{"text":" world"}]},"finishReason":"STOP"}],"usageMetadata":{"promptTokenCount":100,"candidatesTokenCount":30,"thoughtsTokenCount":80,"cachedContentTokenCount":10}}`)
+		fmt.Fprintln(pw, "")
+	}()
+
+	result, err := svc.handleGeminiStreamingResponse(c, resp, time.Now())
+	_ = pr.Close()
+
+	require.NoError(t, err)
+	require.NotNil(t, result)
+	require.NotNil(t, result.usage)
+	// promptTokenCount=100, cachedContentTokenCount=10 → InputTokens=90
+	require.Equal(t, 90, result.usage.InputTokens)
+	// candidatesTokenCount=30 + thoughtsTokenCount=80 → OutputTokens=110
+	require.Equal(t, 110, result.usage.OutputTokens)
+	require.Equal(t, 10, result.usage.CacheReadInputTokens)
+}
+
+// TestHandleClaudeStreamingResponse_ThoughtsTokenCount
+// 验证：Gemini→Claude 流式转换时 thoughtsTokenCount 被计入 OutputTokens
+func TestHandleClaudeStreamingResponse_ThoughtsTokenCount(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	svc := newAntigravityTestService(&config.Config{
+		Gateway: config.GatewayConfig{MaxLineSize: defaultMaxLineSize},
+	})
+
+	rec := httptest.NewRecorder()
+	c, _ := gin.CreateTestContext(rec)
+	c.Request = httptest.NewRequest(http.MethodPost, "/", nil)
+
+	pr, pw := io.Pipe()
+	resp := &http.Response{StatusCode: http.StatusOK, Body: pr, Header: http.Header{}}
+
+	go func() {
+		defer func() { _ = pw.Close() }()
+		fmt.Fprintln(pw, `data: {"response":{"candidates":[{"content":{"parts":[{"text":"Hi"}]},"finishReason":"STOP"}],"usageMetadata":{"promptTokenCount":50,"candidatesTokenCount":10,"thoughtsTokenCount":25}}}`)
+		fmt.Fprintln(pw, "")
+	}()
+
+	result, err := svc.handleClaudeStreamingResponse(c, resp, time.Now(), "gemini-2.5-pro")
+	_ = pr.Close()
+
+	require.NoError(t, err)
+	require.NotNil(t, result)
+	require.NotNil(t, result.usage)
+	// promptTokenCount=50 → InputTokens=50
+	require.Equal(t, 50, result.usage.InputTokens)
+	// candidatesTokenCount=10 + thoughtsTokenCount=25 → OutputTokens=35
+	require.Equal(t, 35, result.usage.OutputTokens)
+}
+
 // --- 流式客户端断开检测测试 ---
 
 // TestStreamUpstreamResponse_ClientDisconnectDrainsUsage

backend/internal/service/antigravity_oauth_service.go

@@ -192,6 +192,43 @@ func (s *AntigravityOAuthService) RefreshToken(ctx context.Context, refreshToken
 	return nil, fmt.Errorf("token 刷新失败 (重试后): %w", lastErr)
 }
 
+// ValidateRefreshToken 用 refresh token 验证并获取完整的 token 信息（含 email 和 project_id）
+func (s *AntigravityOAuthService) ValidateRefreshToken(ctx context.Context, refreshToken string, proxyID *int64) (*AntigravityTokenInfo, error) {
+	var proxyURL string
+	if proxyID != nil {
+		proxy, err := s.proxyRepo.GetByID(ctx, *proxyID)
+		if err == nil && proxy != nil {
+			proxyURL = proxy.URL()
+		}
+	}
+
+	// 刷新 token
+	tokenInfo, err := s.RefreshToken(ctx, refreshToken, proxyURL)
+	if err != nil {
+		return nil, err
+	}
+
+	// 获取用户信息（email）
+	client := antigravity.NewClient(proxyURL)
+	userInfo, err := client.GetUserInfo(ctx, tokenInfo.AccessToken)
+	if err != nil {
+		fmt.Printf("[AntigravityOAuth] 警告: 获取用户信息失败: %v\n", err)
+	} else {
+		tokenInfo.Email = userInfo.Email
+	}
+
+	// 获取 project_id（容错，失败不阻塞）
+	projectID, loadErr := s.loadProjectIDWithRetry(ctx, tokenInfo.AccessToken, proxyURL, 3)
+	if loadErr != nil {
+		fmt.Printf("[AntigravityOAuth] 警告: 获取 project_id 失败（重试后）: %v\n", loadErr)
+		tokenInfo.ProjectIDMissing = true
+	} else {
+		tokenInfo.ProjectID = projectID
+	}
+
+	return tokenInfo, nil
+}
+
 func isNonRetryableAntigravityOAuthError(err error) bool {
 	msg := err.Error()
 	nonRetryable := []string{

backend/internal/service/billing_service.go

@@ -31,8 +31,8 @@ type ModelPricing struct {
 	OutputPricePerToken        float64 // 每token输出价格 (USD)
 	CacheCreationPricePerToken float64 // 缓存创建每token价格 (USD)
 	CacheReadPricePerToken     float64 // 缓存读取每token价格 (USD)
-	CacheCreation5mPrice       float64 // 5分钟缓存创建价格（每百万token）- 仅用于硬编码回退
+	CacheCreation5mPrice       float64 // 5分钟缓存创建每token价格 (USD)
-	CacheCreation1hPrice       float64 // 1小时缓存创建价格（每百万token）- 仅用于硬编码回退
+	CacheCreation1hPrice       float64 // 1小时缓存创建每token价格 (USD)
 	SupportsCacheBreakdown     bool    // 是否支持详细的缓存分类
 }
 
@@ -172,12 +172,20 @@ func (s *BillingService) GetModelPricing(model string) (*ModelPricing, error) {
 	if s.pricingService != nil {
 		litellmPricing := s.pricingService.GetModelPricing(model)
 		if litellmPricing != nil {
+			// 启用 5m/1h 分类计费的条件：
+			// 1. 存在 1h 价格
+			// 2. 1h 价格 > 5m 价格（防止 LiteLLM 数据错误导致少收费）
+			price5m := litellmPricing.CacheCreationInputTokenCost
+			price1h := litellmPricing.CacheCreationInputTokenCostAbove1hr
+			enableBreakdown := price1h > 0 && price1h > price5m
 			return &ModelPricing{
 				InputPricePerToken:         litellmPricing.InputCostPerToken,
 				OutputPricePerToken:        litellmPricing.OutputCostPerToken,
 				CacheCreationPricePerToken: litellmPricing.CacheCreationInputTokenCost,
 				CacheReadPricePerToken:     litellmPricing.CacheReadInputTokenCost,
-				SupportsCacheBreakdown:     false,
+				CacheCreation5mPrice:       price5m,
+				CacheCreation1hPrice:       price1h,
+				SupportsCacheBreakdown:     enableBreakdown,
 			}, nil
 		}
 	}
@@ -209,9 +217,14 @@ func (s *BillingService) CalculateCost(model string, tokens UsageTokens, rateMul
 
 	// 计算缓存费用
 	if pricing.SupportsCacheBreakdown && (pricing.CacheCreation5mPrice > 0 || pricing.CacheCreation1hPrice > 0) {
-		// 支持详细缓存分类的模型（5分钟/1小时缓存）
+		// 支持详细缓存分类的模型（5分钟/1小时缓存，价格为 per-token）
-		breakdown.CacheCreationCost = float64(tokens.CacheCreation5mTokens)/1_000_000*pricing.CacheCreation5mPrice +
+		if tokens.CacheCreation5mTokens == 0 && tokens.CacheCreation1hTokens == 0 && tokens.CacheCreationTokens > 0 {
-			float64(tokens.CacheCreation1hTokens)/1_000_000*pricing.CacheCreation1hPrice
+			// API 未返回 ephemeral 明细，回退到全部按 5m 单价计费
+			breakdown.CacheCreationCost = float64(tokens.CacheCreationTokens) * pricing.CacheCreation5mPrice
+		} else {
+			breakdown.CacheCreationCost = float64(tokens.CacheCreation5mTokens)*pricing.CacheCreation5mPrice +
+				float64(tokens.CacheCreation1hTokens)*pricing.CacheCreation1hPrice
+		}
 	} else {
 		// 标准缓存创建价格（per-token）
 		breakdown.CacheCreationCost = float64(tokens.CacheCreationTokens) * pricing.CacheCreationPricePerToken
@@ -280,10 +293,12 @@ func (s *BillingService) CalculateCostWithLongContext(model string, tokens Usage
 
 	// 范围内部分：正常计费
 	inRangeTokens := UsageTokens{
-		InputTokens:         inRangeInputTokens,
+		InputTokens:           inRangeInputTokens,
-		OutputTokens:        tokens.OutputTokens, // 输出只算一次
+		OutputTokens:          tokens.OutputTokens, // 输出只算一次
-		CacheCreationTokens: tokens.CacheCreationTokens,
+		CacheCreationTokens:   tokens.CacheCreationTokens,
-		CacheReadTokens:     inRangeCacheTokens,
+		CacheReadTokens:       inRangeCacheTokens,
+		CacheCreation5mTokens: tokens.CacheCreation5mTokens,
+		CacheCreation1hTokens: tokens.CacheCreation1hTokens,
 	}
 	inRangeCost, err := s.CalculateCost(model, inRangeTokens, rateMultiplier)
 	if err != nil {

backend/internal/service/gateway_beta_test.go

@@ -21,3 +21,72 @@ func TestMergeAnthropicBeta_EmptyIncoming(t *testing.T) {
 	)
 	require.Equal(t, "oauth-2025-04-20,interleaved-thinking-2025-05-14", got)
 }
+
+func TestStripBetaToken(t *testing.T) {
+	tests := []struct {
+		name   string
+		header string
+		token  string
+		want   string
+	}{
+		{
+			name:   "token in middle",
+			header: "oauth-2025-04-20,context-1m-2025-08-07,interleaved-thinking-2025-05-14",
+			token:  "context-1m-2025-08-07",
+			want:   "oauth-2025-04-20,interleaved-thinking-2025-05-14",
+		},
+		{
+			name:   "token at start",
+			header: "context-1m-2025-08-07,oauth-2025-04-20,interleaved-thinking-2025-05-14",
+			token:  "context-1m-2025-08-07",
+			want:   "oauth-2025-04-20,interleaved-thinking-2025-05-14",
+		},
+		{
+			name:   "token at end",
+			header: "oauth-2025-04-20,interleaved-thinking-2025-05-14,context-1m-2025-08-07",
+			token:  "context-1m-2025-08-07",
+			want:   "oauth-2025-04-20,interleaved-thinking-2025-05-14",
+		},
+		{
+			name:   "token not present",
+			header: "oauth-2025-04-20,interleaved-thinking-2025-05-14",
+			token:  "context-1m-2025-08-07",
+			want:   "oauth-2025-04-20,interleaved-thinking-2025-05-14",
+		},
+		{
+			name:   "empty header",
+			header: "",
+			token:  "context-1m-2025-08-07",
+			want:   "",
+		},
+		{
+			name:   "with spaces",
+			header: "oauth-2025-04-20, context-1m-2025-08-07 , interleaved-thinking-2025-05-14",
+			token:  "context-1m-2025-08-07",
+			want:   "oauth-2025-04-20,interleaved-thinking-2025-05-14",
+		},
+		{
+			name:   "only token",
+			header: "context-1m-2025-08-07",
+			token:  "context-1m-2025-08-07",
+			want:   "",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			got := stripBetaToken(tt.header, tt.token)
+			require.Equal(t, tt.want, got)
+		})
+	}
+}
+
+func TestMergeAnthropicBetaDropping_Context1M(t *testing.T) {
+	required := []string{"oauth-2025-04-20", "interleaved-thinking-2025-05-14"}
+	incoming := "context-1m-2025-08-07,foo-beta,oauth-2025-04-20"
+	drop := map[string]struct{}{"context-1m-2025-08-07": {}}
+
+	got := mergeAnthropicBetaDropping(required, incoming, drop)
+	require.Equal(t, "oauth-2025-04-20,interleaved-thinking-2025-05-14,foo-beta", got)
+	require.NotContains(t, got, "context-1m-2025-08-07")
+}

backend/internal/service/gateway_multiplatform_test.go

@@ -87,7 +87,7 @@ func (m *mockAccountRepoForPlatform) Delete(ctx context.Context, id int64) error
 func (m *mockAccountRepoForPlatform) List(ctx context.Context, params pagination.PaginationParams) ([]Account, *pagination.PaginationResult, error) {
 	return nil, nil, nil
 }
-func (m *mockAccountRepoForPlatform) ListWithFilters(ctx context.Context, params pagination.PaginationParams, platform, accountType, status, search string) ([]Account, *pagination.PaginationResult, error) {
+func (m *mockAccountRepoForPlatform) ListWithFilters(ctx context.Context, params pagination.PaginationParams, platform, accountType, status, search string, groupID int64) ([]Account, *pagination.PaginationResult, error) {
 	return nil, nil, nil
 }
 func (m *mockAccountRepoForPlatform) ListByGroup(ctx context.Context, groupID int64) ([]Account, error) {

backend/internal/service/gateway_request.go

@@ -101,9 +101,9 @@ func ParseGatewayRequest(body []byte, protocol string) (*ParsedRequest, error) {
 		}
 	}
 
-	// thinking: {type: "enabled"}
+	// thinking: {type: "enabled" | "adaptive"}
 	if rawThinking, ok := req["thinking"].(map[string]any); ok {
-		if t, ok := rawThinking["type"].(string); ok && t == "enabled" {
+		if t, ok := rawThinking["type"].(string); ok && (t == "enabled" || t == "adaptive") {
 			parsed.ThinkingEnabled = true
 		}
 	}
@@ -161,9 +161,9 @@ func parseIntegralNumber(raw any) (int, bool) {
 // Returns filtered body or original body if filtering fails (fail-safe)
 // This prevents 400 errors from invalid thinking block signatures
 //
-// Strategy:
+// 策略：
-//   - When thinking.type != "enabled": Remove all thinking blocks
+//   - 当 thinking.type 不是 "enabled"/"adaptive"：移除所有 thinking 相关块
-//   - When thinking.type == "enabled": Only remove thinking blocks without valid signatures
+//   - 当 thinking.type 是 "enabled"/"adaptive"：仅移除缺失/无效 signature 的 thinking 块（避免 400）
 //     (blocks with missing/empty/dummy signatures that would cause 400 errors)
 func FilterThinkingBlocks(body []byte) []byte {
 	return filterThinkingBlocksInternal(body, false)
@@ -489,9 +489,9 @@ func FilterSignatureSensitiveBlocksForRetry(body []byte) []byte {
 }
 
 // filterThinkingBlocksInternal removes invalid thinking blocks from request
-// Strategy:
+// 策略：
-//   - When thinking.type != "enabled": Remove all thinking blocks
+//   - 当 thinking.type 不是 "enabled"/"adaptive"：移除所有 thinking 相关块
-//   - When thinking.type == "enabled": Only remove thinking blocks without valid signatures
+//   - 当 thinking.type 是 "enabled"/"adaptive"：仅移除缺失/无效 signature 的 thinking 块
 func filterThinkingBlocksInternal(body []byte, _ bool) []byte {
 	// Fast path: if body doesn't contain "thinking", skip parsing
 	if !bytes.Contains(body, []byte(`"type":"thinking"`)) &&
@@ -511,7 +511,7 @@ func filterThinkingBlocksInternal(body []byte, _ bool) []byte {
 	// Check if thinking is enabled
 	thinkingEnabled := false
 	if thinking, ok := req["thinking"].(map[string]any); ok {
-		if thinkType, ok := thinking["type"].(string); ok && thinkType == "enabled" {
+		if thinkType, ok := thinking["type"].(string); ok && (thinkType == "enabled" || thinkType == "adaptive") {
 			thinkingEnabled = true
 		}
 	}

backend/internal/service/gateway_request_test.go

@@ -29,6 +29,14 @@ func TestParseGatewayRequest_ThinkingEnabled(t *testing.T) {
 	require.True(t, parsed.ThinkingEnabled)
 }
 
+func TestParseGatewayRequest_ThinkingAdaptiveEnabled(t *testing.T) {
+	body := []byte(`{"model":"claude-sonnet-4-5","thinking":{"type":"adaptive"},"messages":[{"content":"hi"}]}`)
+	parsed, err := ParseGatewayRequest(body, "")
+	require.NoError(t, err)
+	require.Equal(t, "claude-sonnet-4-5", parsed.Model)
+	require.True(t, parsed.ThinkingEnabled)
+}
+
 func TestParseGatewayRequest_MaxTokens(t *testing.T) {
 	body := []byte(`{"model":"claude-haiku-4-5","max_tokens":1}`)
 	parsed, err := ParseGatewayRequest(body, "")
@@ -209,6 +217,16 @@ func TestFilterThinkingBlocks(t *testing.T) {
 			input:        `{"model":"claude-3-5-sonnet-20241022","messages":[{"role":"user","content":[{"type":"text","text":"Hello"},{"type":"thinking","thinking":"internal","signature":"invalid"},{"type":"text","text":"World"}]}]}`,
 			shouldFilter: true,
 		},
+		{
+			name:         "does not filter signed thinking blocks when thinking adaptive",
+			input:        `{"thinking":{"type":"adaptive"},"messages":[{"role":"assistant","content":[{"type":"thinking","thinking":"ok","signature":"sig_real_123"},{"type":"text","text":"B"}]}]}`,
+			shouldFilter: false,
+		},
+		{
+			name:         "filters unsigned thinking blocks when thinking adaptive",
+			input:        `{"thinking":{"type":"adaptive"},"messages":[{"role":"assistant","content":[{"type":"thinking","thinking":"internal","signature":""},{"type":"text","text":"B"}]}]}`,
+			shouldFilter: true,
+		},
 		{
 			name:         "handles no thinking blocks",
 			input:        `{"model":"claude-3-5-sonnet-20241022","messages":[{"role":"user","content":[{"type":"text","text":"Hello"}]}]}`,

backend/internal/service/gateway_service.go

@@ -349,6 +349,8 @@ type ClaudeUsage struct {
 	OutputTokens             int `json:"output_tokens"`
 	CacheCreationInputTokens int `json:"cache_creation_input_tokens"`
 	CacheReadInputTokens     int `json:"cache_read_input_tokens"`
+	CacheCreation5mTokens    int // 5分钟缓存创建token（来自嵌套 cache_creation 对象）
+	CacheCreation1hTokens    int // 1小时缓存创建token（来自嵌套 cache_creation 对象）
 }
 
 // ForwardResult 转发结果
@@ -3551,12 +3553,12 @@ func (s *GatewayService) buildUpstreamRequest(ctx context.Context, c *gin.Contex
 			// messages requests typically use only oauth + interleaved-thinking.
 			// Also drop claude-code beta if a downstream client added it.
 			requiredBetas := []string{claude.BetaOAuth, claude.BetaInterleavedThinking}
-			drop := map[string]struct{}{claude.BetaClaudeCode: {}}
+			drop := map[string]struct{}{claude.BetaClaudeCode: {}, claude.BetaContext1M: {}}
 			req.Header.Set("anthropic-beta", mergeAnthropicBetaDropping(requiredBetas, incomingBeta, drop))
 		} else {
 			// Claude Code 客户端：尽量透传原始 header，仅补齐 oauth beta
 			clientBetaHeader := req.Header.Get("anthropic-beta")
-			req.Header.Set("anthropic-beta", s.getBetaHeader(modelID, clientBetaHeader))
+			req.Header.Set("anthropic-beta", stripBetaToken(s.getBetaHeader(modelID, clientBetaHeader), claude.BetaContext1M))
 		}
 	} else if s.cfg != nil && s.cfg.Gateway.InjectBetaForAPIKey && req.Header.Get("anthropic-beta") == "" {
 		// API-key：仅在请求显式使用 beta 特性且客户端未提供时，按需补齐（默认关闭）
@@ -3631,7 +3633,8 @@ func requestNeedsBetaFeatures(body []byte) bool {
 	if tools.Exists() && tools.IsArray() && len(tools.Array()) > 0 {
 		return true
 	}
-	if strings.EqualFold(gjson.GetBytes(body, "thinking.type").String(), "enabled") {
+	thinkingType := gjson.GetBytes(body, "thinking.type").String()
+	if strings.EqualFold(thinkingType, "enabled") || strings.EqualFold(thinkingType, "adaptive") {
 		return true
 	}
 	return false
@@ -3709,6 +3712,23 @@ func mergeAnthropicBetaDropping(required []string, incoming string, drop map[str
 	return strings.Join(out, ",")
 }
 
+// stripBetaToken removes a single beta token from a comma-separated header value.
+// It short-circuits when the token is not present to avoid unnecessary allocations.
+func stripBetaToken(header, token string) string {
+	if !strings.Contains(header, token) {
+		return header
+	}
+	out := make([]string, 0, 8)
+	for _, p := range strings.Split(header, ",") {
+		p = strings.TrimSpace(p)
+		if p == "" || p == token {
+			continue
+		}
+		out = append(out, p)
+	}
+	return strings.Join(out, ",")
+}
+
 // applyClaudeCodeMimicHeaders forces "Claude Code-like" request headers.
 // This mirrors opencode-anthropic-auth behavior: do not trust downstream
 // headers when using Claude Code-scoped OAuth credentials.
@@ -4273,6 +4293,23 @@ func (s *GatewayService) handleStreamingResponse(ctx context.Context, resp *http
 			}
 		}
 
+		// Cache TTL Override: 重写 SSE 事件中的 cache_creation 分类
+		if account.IsCacheTTLOverrideEnabled() {
+			overrideTarget := account.GetCacheTTLOverrideTarget()
+			if eventType == "message_start" {
+				if msg, ok := event["message"].(map[string]any); ok {
+					if u, ok := msg["usage"].(map[string]any); ok {
+						rewriteCacheCreationJSON(u, overrideTarget)
+					}
+				}
+			}
+			if eventType == "message_delta" {
+				if u, ok := event["usage"].(map[string]any); ok {
+					rewriteCacheCreationJSON(u, overrideTarget)
+				}
+			}
+		}
+
 		if needModelReplace {
 			if msg, ok := event["message"].(map[string]any); ok {
 				if model, ok := msg["model"].(string); ok && model == mappedModel {
@@ -4400,6 +4437,14 @@ func (s *GatewayService) parseSSEUsage(data string, usage *ClaudeUsage) {
 		usage.InputTokens = msgStart.Message.Usage.InputTokens
 		usage.CacheCreationInputTokens = msgStart.Message.Usage.CacheCreationInputTokens
 		usage.CacheReadInputTokens = msgStart.Message.Usage.CacheReadInputTokens
+
+		// 解析嵌套的 cache_creation 对象中的 5m/1h 明细
+		cc5m := gjson.Get(data, "message.usage.cache_creation.ephemeral_5m_input_tokens")
+		cc1h := gjson.Get(data, "message.usage.cache_creation.ephemeral_1h_input_tokens")
+		if cc5m.Exists() || cc1h.Exists() {
+			usage.CacheCreation5mTokens = int(cc5m.Int())
+			usage.CacheCreation1hTokens = int(cc1h.Int())
+		}
 	}
 
 	// 解析message_delta获取tokens（兼容GLM等把所有usage放在delta中的API）
@@ -4428,6 +4473,66 @@ func (s *GatewayService) parseSSEUsage(data string, usage *ClaudeUsage) {
 		if msgDelta.Usage.CacheReadInputTokens > 0 {
 			usage.CacheReadInputTokens = msgDelta.Usage.CacheReadInputTokens
 		}
+
+		// 解析嵌套的 cache_creation 对象中的 5m/1h 明细
+		cc5m := gjson.Get(data, "usage.cache_creation.ephemeral_5m_input_tokens")
+		cc1h := gjson.Get(data, "usage.cache_creation.ephemeral_1h_input_tokens")
+		if cc5m.Exists() || cc1h.Exists() {
+			usage.CacheCreation5mTokens = int(cc5m.Int())
+			usage.CacheCreation1hTokens = int(cc1h.Int())
+		}
+	}
+}
+
+// applyCacheTTLOverride 将所有 cache creation tokens 归入指定的 TTL 类型。
+// target 为 "5m" 或 "1h"。返回 true 表示发生了变更。
+func applyCacheTTLOverride(usage *ClaudeUsage, target string) bool {
+	// Fallback: 如果只有聚合字段但无 5m/1h 明细，将聚合字段归入 5m 默认类别
+	if usage.CacheCreation5mTokens == 0 && usage.CacheCreation1hTokens == 0 && usage.CacheCreationInputTokens > 0 {
+		usage.CacheCreation5mTokens = usage.CacheCreationInputTokens
+	}
+
+	total := usage.CacheCreation5mTokens + usage.CacheCreation1hTokens
+	if total == 0 {
+		return false
+	}
+	switch target {
+	case "1h":
+		if usage.CacheCreation1hTokens == total {
+			return false // 已经全是 1h
+		}
+		usage.CacheCreation1hTokens = total
+		usage.CacheCreation5mTokens = 0
+	default: // "5m"
+		if usage.CacheCreation5mTokens == total {
+			return false // 已经全是 5m
+		}
+		usage.CacheCreation5mTokens = total
+		usage.CacheCreation1hTokens = 0
+	}
+	return true
+}
+
+// rewriteCacheCreationJSON 在 JSON usage 对象中重写 cache_creation 嵌套对象的 TTL 分类。
+// usageObj 是 usage JSON 对象（map[string]any）。
+func rewriteCacheCreationJSON(usageObj map[string]any, target string) {
+	ccObj, ok := usageObj["cache_creation"].(map[string]any)
+	if !ok {
+		return
+	}
+	v5m, _ := ccObj["ephemeral_5m_input_tokens"].(float64)
+	v1h, _ := ccObj["ephemeral_1h_input_tokens"].(float64)
+	total := v5m + v1h
+	if total == 0 {
+		return
+	}
+	switch target {
+	case "1h":
+		ccObj["ephemeral_1h_input_tokens"] = total
+		ccObj["ephemeral_5m_input_tokens"] = float64(0)
+	default: // "5m"
+		ccObj["ephemeral_5m_input_tokens"] = total
+		ccObj["ephemeral_1h_input_tokens"] = float64(0)
 	}
 }
 
@@ -4448,6 +4553,14 @@ func (s *GatewayService) handleNonStreamingResponse(ctx context.Context, resp *h
 		return nil, fmt.Errorf("parse response: %w", err)
 	}
 
+	// 解析嵌套的 cache_creation 对象中的 5m/1h 明细
+	cc5m := gjson.GetBytes(body, "usage.cache_creation.ephemeral_5m_input_tokens")
+	cc1h := gjson.GetBytes(body, "usage.cache_creation.ephemeral_1h_input_tokens")
+	if cc5m.Exists() || cc1h.Exists() {
+		response.Usage.CacheCreation5mTokens = int(cc5m.Int())
+		response.Usage.CacheCreation1hTokens = int(cc1h.Int())
+	}
+
 	// 兼容 Kimi cached_tokens → cache_read_input_tokens
 	if response.Usage.CacheReadInputTokens == 0 {
 		cachedTokens := gjson.GetBytes(body, "usage.cached_tokens").Int()
@@ -4459,6 +4572,20 @@ func (s *GatewayService) handleNonStreamingResponse(ctx context.Context, resp *h
 		}
 	}
 
+	// Cache TTL Override: 重写 non-streaming 响应中的 cache_creation 分类
+	if account.IsCacheTTLOverrideEnabled() {
+		overrideTarget := account.GetCacheTTLOverrideTarget()
+		if applyCacheTTLOverride(&response.Usage, overrideTarget) {
+			// 同步更新 body JSON 中的嵌套 cache_creation 对象
+			if newBody, err := sjson.SetBytes(body, "usage.cache_creation.ephemeral_5m_input_tokens", response.Usage.CacheCreation5mTokens); err == nil {
+				body = newBody
+			}
+			if newBody, err := sjson.SetBytes(body, "usage.cache_creation.ephemeral_1h_input_tokens", response.Usage.CacheCreation1hTokens); err == nil {
+				body = newBody
+			}
+		}
+	}
+
 	// 如果有模型映射，替换响应中的model字段
 	if originalModel != mappedModel {
 		body = s.replaceModelInResponseBody(body, mappedModel, originalModel)
@@ -4535,6 +4662,13 @@ func (s *GatewayService) RecordUsage(ctx context.Context, input *RecordUsageInpu
 		result.Usage.InputTokens = 0
 	}
 
+	// Cache TTL Override: 确保计费时 token 分类与账号设置一致
+	cacheTTLOverridden := false
+	if account.IsCacheTTLOverrideEnabled() {
+		applyCacheTTLOverride(&result.Usage, account.GetCacheTTLOverrideTarget())
+		cacheTTLOverridden = (result.Usage.CacheCreation5mTokens + result.Usage.CacheCreation1hTokens) > 0
+	}
+
 	// 获取费率倍数（优先级：用户专属 > 分组默认 > 系统默认）
 	multiplier := s.cfg.Default.RateMultiplier
 	if apiKey.GroupID != nil && apiKey.Group != nil {
@@ -4565,10 +4699,12 @@ func (s *GatewayService) RecordUsage(ctx context.Context, input *RecordUsageInpu
 	} else {
 		// Token 计费
 		tokens := UsageTokens{
-			InputTokens:         result.Usage.InputTokens,
+			InputTokens:           result.Usage.InputTokens,
-			OutputTokens:        result.Usage.OutputTokens,
+			OutputTokens:          result.Usage.OutputTokens,
-			CacheCreationTokens: result.Usage.CacheCreationInputTokens,
+			CacheCreationTokens:   result.Usage.CacheCreationInputTokens,
-			CacheReadTokens:     result.Usage.CacheReadInputTokens,
+			CacheReadTokens:       result.Usage.CacheReadInputTokens,
+			CacheCreation5mTokens: result.Usage.CacheCreation5mTokens,
+			CacheCreation1hTokens: result.Usage.CacheCreation1hTokens,
 		}
 		var err error
 		cost, err = s.billingService.CalculateCost(result.Model, tokens, multiplier)
@@ -4602,6 +4738,8 @@ func (s *GatewayService) RecordUsage(ctx context.Context, input *RecordUsageInpu
 		OutputTokens:          result.Usage.OutputTokens,
 		CacheCreationTokens:   result.Usage.CacheCreationInputTokens,
 		CacheReadTokens:       result.Usage.CacheReadInputTokens,
+		CacheCreation5mTokens: result.Usage.CacheCreation5mTokens,
+		CacheCreation1hTokens: result.Usage.CacheCreation1hTokens,
 		InputCost:             cost.InputCost,
 		OutputCost:            cost.OutputCost,
 		CacheCreationCost:     cost.CacheCreationCost,
@@ -4616,6 +4754,7 @@ func (s *GatewayService) RecordUsage(ctx context.Context, input *RecordUsageInpu
 		FirstTokenMs:          result.FirstTokenMs,
 		ImageCount:            result.ImageCount,
 		ImageSize:             imageSize,
+		CacheTTLOverridden:    cacheTTLOverridden,
 		CreatedAt:             time.Now(),
 	}
 
@@ -4716,6 +4855,13 @@ func (s *GatewayService) RecordUsageWithLongContext(ctx context.Context, input *
 		result.Usage.InputTokens = 0
 	}
 
+	// Cache TTL Override: 确保计费时 token 分类与账号设置一致
+	cacheTTLOverridden := false
+	if account.IsCacheTTLOverrideEnabled() {
+		applyCacheTTLOverride(&result.Usage, account.GetCacheTTLOverrideTarget())
+		cacheTTLOverridden = (result.Usage.CacheCreation5mTokens + result.Usage.CacheCreation1hTokens) > 0
+	}
+
 	// 获取费率倍数（优先级：用户专属 > 分组默认 > 系统默认）
 	multiplier := s.cfg.Default.RateMultiplier
 	if apiKey.GroupID != nil && apiKey.Group != nil {
@@ -4746,10 +4892,12 @@ func (s *GatewayService) RecordUsageWithLongContext(ctx context.Context, input *
 	} else {
 		// Token 计费（使用长上下文计费方法）
 		tokens := UsageTokens{
-			InputTokens:         result.Usage.InputTokens,
+			InputTokens:           result.Usage.InputTokens,
-			OutputTokens:        result.Usage.OutputTokens,
+			OutputTokens:          result.Usage.OutputTokens,
-			CacheCreationTokens: result.Usage.CacheCreationInputTokens,
+			CacheCreationTokens:   result.Usage.CacheCreationInputTokens,
-			CacheReadTokens:     result.Usage.CacheReadInputTokens,
+			CacheReadTokens:       result.Usage.CacheReadInputTokens,
+			CacheCreation5mTokens: result.Usage.CacheCreation5mTokens,
+			CacheCreation1hTokens: result.Usage.CacheCreation1hTokens,
 		}
 		var err error
 		cost, err = s.billingService.CalculateCostWithLongContext(result.Model, tokens, multiplier, input.LongContextThreshold, input.LongContextMultiplier)
@@ -4783,6 +4931,8 @@ func (s *GatewayService) RecordUsageWithLongContext(ctx context.Context, input *
 		OutputTokens:          result.Usage.OutputTokens,
 		CacheCreationTokens:   result.Usage.CacheCreationInputTokens,
 		CacheReadTokens:       result.Usage.CacheReadInputTokens,
+		CacheCreation5mTokens: result.Usage.CacheCreation5mTokens,
+		CacheCreation1hTokens: result.Usage.CacheCreation1hTokens,
 		InputCost:             cost.InputCost,
 		OutputCost:            cost.OutputCost,
 		CacheCreationCost:     cost.CacheCreationCost,
@@ -4797,6 +4947,7 @@ func (s *GatewayService) RecordUsageWithLongContext(ctx context.Context, input *
 		FirstTokenMs:          result.FirstTokenMs,
 		ImageCount:            result.ImageCount,
 		ImageSize:             imageSize,
+		CacheTTLOverridden:    cacheTTLOverridden,
 		CreatedAt:             time.Now(),
 	}
 
@@ -5102,7 +5253,8 @@ func (s *GatewayService) buildCountTokensRequest(ctx context.Context, c *gin.Con
 
 			incomingBeta := req.Header.Get("anthropic-beta")
 			requiredBetas := []string{claude.BetaClaudeCode, claude.BetaOAuth, claude.BetaInterleavedThinking, claude.BetaTokenCounting}
-			req.Header.Set("anthropic-beta", mergeAnthropicBeta(requiredBetas, incomingBeta))
+			drop := map[string]struct{}{claude.BetaContext1M: {}}
+			req.Header.Set("anthropic-beta", mergeAnthropicBetaDropping(requiredBetas, incomingBeta, drop))
 		} else {
 			clientBetaHeader := req.Header.Get("anthropic-beta")
 			if clientBetaHeader == "" {
@@ -5112,7 +5264,7 @@ func (s *GatewayService) buildCountTokensRequest(ctx context.Context, c *gin.Con
 				if !strings.Contains(beta, claude.BetaTokenCounting) {
 					beta = beta + "," + claude.BetaTokenCounting
 				}
-				req.Header.Set("anthropic-beta", beta)
+				req.Header.Set("anthropic-beta", stripBetaToken(beta, claude.BetaContext1M))
 			}
 		}
 	} else if s.cfg != nil && s.cfg.Gateway.InjectBetaForAPIKey && req.Header.Get("anthropic-beta") == "" {

backend/internal/service/gemini_messages_compat_service.go

@@ -2663,11 +2663,12 @@ func extractGeminiUsage(geminiResp map[string]any) *ClaudeUsage {
 	prompt, _ := asInt(usageMeta["promptTokenCount"])
 	cand, _ := asInt(usageMeta["candidatesTokenCount"])
 	cached, _ := asInt(usageMeta["cachedContentTokenCount"])
+	thoughts, _ := asInt(usageMeta["thoughtsTokenCount"])
 	// 注意：Gemini 的 promptTokenCount 包含 cachedContentTokenCount，
 	// 但 Claude 的 input_tokens 不包含 cache_read_input_tokens，需要减去
 	return &ClaudeUsage{
 		InputTokens:          prompt - cached,
-		OutputTokens:         cand,
+		OutputTokens:         cand + thoughts,
 		CacheReadInputTokens: cached,
 	}
 }

backend/internal/service/gemini_messages_compat_service_test.go

@@ -4,6 +4,8 @@ import (
 	"encoding/json"
 	"strings"
 	"testing"
+
+	"github.com/stretchr/testify/require"
 )
 
 // TestConvertClaudeToolsToGeminiTools_CustomType 测试custom类型工具转换
@@ -203,3 +205,70 @@ func TestEnsureGeminiFunctionCallThoughtSignatures_InsertsWhenMissing(t *testing
 		t.Fatalf("expected injected thoughtSignature %q, got: %s", geminiDummyThoughtSignature, s)
 	}
 }
+
+func TestExtractGeminiUsage_ThoughtsTokenCount(t *testing.T) {
+	tests := []struct {
+		name          string
+		resp          map[string]any
+		wantInput     int
+		wantOutput    int
+		wantCacheRead int
+		wantNil       bool
+	}{
+		{
+			name: "with thoughtsTokenCount",
+			resp: map[string]any{
+				"usageMetadata": map[string]any{
+					"promptTokenCount":     float64(100),
+					"candidatesTokenCount": float64(20),
+					"thoughtsTokenCount":   float64(50),
+				},
+			},
+			wantInput:  100,
+			wantOutput: 70,
+		},
+		{
+			name: "with thoughtsTokenCount and cache",
+			resp: map[string]any{
+				"usageMetadata": map[string]any{
+					"promptTokenCount":        float64(100),
+					"candidatesTokenCount":    float64(20),
+					"cachedContentTokenCount": float64(30),
+					"thoughtsTokenCount":      float64(50),
+				},
+			},
+			wantInput:     70,
+			wantOutput:    70,
+			wantCacheRead: 30,
+		},
+		{
+			name: "without thoughtsTokenCount (old model)",
+			resp: map[string]any{
+				"usageMetadata": map[string]any{
+					"promptTokenCount":     float64(100),
+					"candidatesTokenCount": float64(20),
+				},
+			},
+			wantInput:  100,
+			wantOutput: 20,
+		},
+		{
+			name:    "no usageMetadata",
+			resp:    map[string]any{},
+			wantNil: true,
+		},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			usage := extractGeminiUsage(tt.resp)
+			if tt.wantNil {
+				require.Nil(t, usage)
+				return
+			}
+			require.NotNil(t, usage)
+			require.Equal(t, tt.wantInput, usage.InputTokens)
+			require.Equal(t, tt.wantOutput, usage.OutputTokens)
+			require.Equal(t, tt.wantCacheRead, usage.CacheReadInputTokens)
+		})
+	}
+}

backend/internal/service/gemini_multiplatform_test.go

@@ -74,7 +74,7 @@ func (m *mockAccountRepoForGemini) Delete(ctx context.Context, id int64) error
 func (m *mockAccountRepoForGemini) List(ctx context.Context, params pagination.PaginationParams) ([]Account, *pagination.PaginationResult, error) {
 	return nil, nil, nil
 }
-func (m *mockAccountRepoForGemini) ListWithFilters(ctx context.Context, params pagination.PaginationParams, platform, accountType, status, search string) ([]Account, *pagination.PaginationResult, error) {
+func (m *mockAccountRepoForGemini) ListWithFilters(ctx context.Context, params pagination.PaginationParams, platform, accountType, status, search string, groupID int64) ([]Account, *pagination.PaginationResult, error) {
 	return nil, nil, nil
 }
 func (m *mockAccountRepoForGemini) ListByGroup(ctx context.Context, groupID int64) ([]Account, error) {

backend/internal/service/openai_codex_transform.go

@@ -112,13 +112,19 @@ func applyCodexOAuthTransform(reqBody map[string]any, isCodexCLI bool) codexTran
 		result.Modified = true
 	}
 
-	if _, ok := reqBody["max_output_tokens"]; ok {
+	// Strip parameters unsupported by codex models via the Responses API.
-		delete(reqBody, "max_output_tokens")
+	for _, key := range []string{
-		result.Modified = true
+		"max_output_tokens",
-	}
+		"max_completion_tokens",
-	if _, ok := reqBody["max_completion_tokens"]; ok {
+		"temperature",
-		delete(reqBody, "max_completion_tokens")
+		"top_p",
-		result.Modified = true
+		"frequency_penalty",
+		"presence_penalty",
+	} {
+		if _, ok := reqBody[key]; ok {
+			delete(reqBody, key)
+			result.Modified = true
+		}
 	}
 
 	if normalizeCodexTools(reqBody) {

backend/internal/service/ops_concurrency.go

@@ -24,7 +24,7 @@ func (s *OpsService) listAllAccountsForOps(ctx context.Context, platformFilter s
 		accounts, pageInfo, err := s.accountRepo.ListWithFilters(ctx, pagination.PaginationParams{
 			Page:     page,
 			PageSize: opsAccountsPageSize,
-		}, platformFilter, "", "", "")
+		}, platformFilter, "", "", "", 0)
 		if err != nil {
 			return nil, err
 		}

backend/internal/service/pricing_service.go

@@ -27,14 +27,15 @@ var (
 // LiteLLMModelPricing LiteLLM价格数据结构
 // 只保留我们需要的字段，使用指针来处理可能缺失的值
 type LiteLLMModelPricing struct {
-	InputCostPerToken           float64 `json:"input_cost_per_token"`
+	InputCostPerToken                   float64 `json:"input_cost_per_token"`
-	OutputCostPerToken          float64 `json:"output_cost_per_token"`
+	OutputCostPerToken                  float64 `json:"output_cost_per_token"`
-	CacheCreationInputTokenCost float64 `json:"cache_creation_input_token_cost"`
+	CacheCreationInputTokenCost         float64 `json:"cache_creation_input_token_cost"`
-	CacheReadInputTokenCost     float64 `json:"cache_read_input_token_cost"`
+	CacheCreationInputTokenCostAbove1hr float64 `json:"cache_creation_input_token_cost_above_1hr"`
-	LiteLLMProvider             string  `json:"litellm_provider"`
+	CacheReadInputTokenCost             float64 `json:"cache_read_input_token_cost"`
-	Mode                        string  `json:"mode"`
+	LiteLLMProvider                     string  `json:"litellm_provider"`
-	SupportsPromptCaching       bool    `json:"supports_prompt_caching"`
+	Mode                                string  `json:"mode"`
-	OutputCostPerImage          float64 `json:"output_cost_per_image"` // 图片生成模型每张图片价格
+	SupportsPromptCaching               bool    `json:"supports_prompt_caching"`
+	OutputCostPerImage                  float64 `json:"output_cost_per_image"` // 图片生成模型每张图片价格
 }
 
 // PricingRemoteClient 远程价格数据获取接口
@@ -45,14 +46,15 @@ type PricingRemoteClient interface {
 
 // LiteLLMRawEntry 用于解析原始JSON数据
 type LiteLLMRawEntry struct {
-	InputCostPerToken           *float64 `json:"input_cost_per_token"`
+	InputCostPerToken                   *float64 `json:"input_cost_per_token"`
-	OutputCostPerToken          *float64 `json:"output_cost_per_token"`
+	OutputCostPerToken                  *float64 `json:"output_cost_per_token"`
-	CacheCreationInputTokenCost *float64 `json:"cache_creation_input_token_cost"`
+	CacheCreationInputTokenCost         *float64 `json:"cache_creation_input_token_cost"`
-	CacheReadInputTokenCost     *float64 `json:"cache_read_input_token_cost"`
+	CacheCreationInputTokenCostAbove1hr *float64 `json:"cache_creation_input_token_cost_above_1hr"`
-	LiteLLMProvider             string   `json:"litellm_provider"`
+	CacheReadInputTokenCost             *float64 `json:"cache_read_input_token_cost"`
-	Mode                        string   `json:"mode"`
+	LiteLLMProvider                     string   `json:"litellm_provider"`
-	SupportsPromptCaching       bool     `json:"supports_prompt_caching"`
+	Mode                                string   `json:"mode"`
-	OutputCostPerImage          *float64 `json:"output_cost_per_image"`
+	SupportsPromptCaching               bool     `json:"supports_prompt_caching"`
+	OutputCostPerImage                  *float64 `json:"output_cost_per_image"`
 }
 
 // PricingService 动态价格服务
@@ -318,6 +320,9 @@ func (s *PricingService) parsePricingData(body []byte) (map[string]*LiteLLMModel
 		if entry.CacheCreationInputTokenCost != nil {
 			pricing.CacheCreationInputTokenCost = *entry.CacheCreationInputTokenCost
 		}
+		if entry.CacheCreationInputTokenCostAbove1hr != nil {
+			pricing.CacheCreationInputTokenCostAbove1hr = *entry.CacheCreationInputTokenCostAbove1hr
+		}
 		if entry.CacheReadInputTokenCost != nil {
 			pricing.CacheReadInputTokenCost = *entry.CacheReadInputTokenCost
 		}

backend/internal/service/ratelimit_service.go

@@ -381,10 +381,31 @@ func (s *RateLimitService) handle429(ctx context.Context, account *Account, head
 		}
 	}
 
-	// 2. 尝试从响应头解析重置时间（Anthropic）
+	// 2. Anthropic 平台：尝试解析 per-window 头（5h / 7d），选择实际触发的窗口
+	if result := calculateAnthropic429ResetTime(headers); result != nil {
+		if err := s.accountRepo.SetRateLimited(ctx, account.ID, result.resetAt); err != nil {
+			slog.Warn("rate_limit_set_failed", "account_id", account.ID, "error", err)
+			return
+		}
+
+		// 更新 session window：优先使用 5h-reset 头精确计算，否则从 resetAt 反推
+		windowEnd := result.resetAt
+		if result.fiveHourReset != nil {
+			windowEnd = *result.fiveHourReset
+		}
+		windowStart := windowEnd.Add(-5 * time.Hour)
+		if err := s.accountRepo.UpdateSessionWindow(ctx, account.ID, &windowStart, &windowEnd, "rejected"); err != nil {
+			slog.Warn("rate_limit_update_session_window_failed", "account_id", account.ID, "error", err)
+		}
+
+		slog.Info("anthropic_account_rate_limited", "account_id", account.ID, "reset_at", result.resetAt, "reset_in", time.Until(result.resetAt).Truncate(time.Second))
+		return
+	}
+
+	// 3. 尝试从响应头解析重置时间（Anthropic 聚合头，向后兼容）
 	resetTimestamp := headers.Get("anthropic-ratelimit-unified-reset")
 
-	// 3. 如果响应头没有，尝试从响应体解析（OpenAI usage_limit_reached, Gemini）
+	// 4. 如果响应头没有，尝试从响应体解析（OpenAI usage_limit_reached, Gemini）
 	if resetTimestamp == "" {
 		switch account.Platform {
 		case PlatformOpenAI:
@@ -497,6 +518,112 @@ func (s *RateLimitService) calculateOpenAI429ResetTime(headers http.Header) *tim
 	return nil
 }
 
+// anthropic429Result holds the parsed Anthropic 429 rate-limit information.
+type anthropic429Result struct {
+	resetAt       time.Time  // The correct reset time to use for SetRateLimited
+	fiveHourReset *time.Time // 5h window reset timestamp (for session window calculation), nil if not available
+}
+
+// calculateAnthropic429ResetTime parses Anthropic's per-window rate-limit headers
+// to determine which window (5h or 7d) actually triggered the 429.
+//
+// Headers used:
+//   - anthropic-ratelimit-unified-5h-utilization / anthropic-ratelimit-unified-5h-surpassed-threshold
+//   - anthropic-ratelimit-unified-5h-reset
+//   - anthropic-ratelimit-unified-7d-utilization / anthropic-ratelimit-unified-7d-surpassed-threshold
+//   - anthropic-ratelimit-unified-7d-reset
+//
+// Returns nil when the per-window headers are absent (caller should fall back to
+// the aggregated anthropic-ratelimit-unified-reset header).
+func calculateAnthropic429ResetTime(headers http.Header) *anthropic429Result {
+	reset5hStr := headers.Get("anthropic-ratelimit-unified-5h-reset")
+	reset7dStr := headers.Get("anthropic-ratelimit-unified-7d-reset")
+
+	if reset5hStr == "" && reset7dStr == "" {
+		return nil
+	}
+
+	var reset5h, reset7d *time.Time
+	if ts, err := strconv.ParseInt(reset5hStr, 10, 64); err == nil {
+		t := time.Unix(ts, 0)
+		reset5h = &t
+	}
+	if ts, err := strconv.ParseInt(reset7dStr, 10, 64); err == nil {
+		t := time.Unix(ts, 0)
+		reset7d = &t
+	}
+
+	is5hExceeded := isAnthropicWindowExceeded(headers, "5h")
+	is7dExceeded := isAnthropicWindowExceeded(headers, "7d")
+
+	slog.Info("anthropic_429_window_analysis",
+		"is_5h_exceeded", is5hExceeded,
+		"is_7d_exceeded", is7dExceeded,
+		"reset_5h", reset5hStr,
+		"reset_7d", reset7dStr,
+	)
+
+	// Select the correct reset time based on which window(s) are exceeded.
+	var chosen *time.Time
+	switch {
+	case is5hExceeded && is7dExceeded:
+		// Both exceeded → prefer 7d (longer cooldown), fall back to 5h
+		chosen = reset7d
+		if chosen == nil {
+			chosen = reset5h
+		}
+	case is5hExceeded:
+		chosen = reset5h
+	case is7dExceeded:
+		chosen = reset7d
+	default:
+		// Neither flag clearly exceeded — pick the sooner reset as best guess
+		chosen = pickSooner(reset5h, reset7d)
+	}
+
+	if chosen == nil {
+		return nil
+	}
+	return &anthropic429Result{resetAt: *chosen, fiveHourReset: reset5h}
+}
+
+// isAnthropicWindowExceeded checks whether a given Anthropic rate-limit window
+// (e.g. "5h" or "7d") has been exceeded, using utilization and surpassed-threshold headers.
+func isAnthropicWindowExceeded(headers http.Header, window string) bool {
+	prefix := "anthropic-ratelimit-unified-" + window + "-"
+
+	// Check surpassed-threshold first (most explicit signal)
+	if st := headers.Get(prefix + "surpassed-threshold"); strings.EqualFold(st, "true") {
+		return true
+	}
+
+	// Fall back to utilization >= 1.0
+	if utilStr := headers.Get(prefix + "utilization"); utilStr != "" {
+		if util, err := strconv.ParseFloat(utilStr, 64); err == nil && util >= 1.0-1e-9 {
+			// Use a small epsilon to handle floating point: treat 0.9999999... as >= 1.0
+			return true
+		}
+	}
+
+	return false
+}
+
+// pickSooner returns whichever of the two time pointers is earlier.
+// If only one is non-nil, it is returned. If both are nil, returns nil.
+func pickSooner(a, b *time.Time) *time.Time {
+	switch {
+	case a != nil && b != nil:
+		if a.Before(*b) {
+			return a
+		}
+		return b
+	case a != nil:
+		return a
+	default:
+		return b
+	}
+}
+
 // parseOpenAIRateLimitResetTime 解析 OpenAI 格式的 429 响应，返回重置时间的 Unix 时间戳
 // OpenAI 的 usage_limit_reached 错误格式：
 //

backend/internal/service/ratelimit_service_anthropic_test.go

@@ -0,0 +1,202 @@
+package service
+
+import (
+	"net/http"
+	"testing"
+	"time"
+)
+
+func TestCalculateAnthropic429ResetTime_Only5hExceeded(t *testing.T) {
+	headers := http.Header{}
+	headers.Set("anthropic-ratelimit-unified-5h-utilization", "1.02")
+	headers.Set("anthropic-ratelimit-unified-5h-reset", "1770998400")
+	headers.Set("anthropic-ratelimit-unified-7d-utilization", "0.32")
+	headers.Set("anthropic-ratelimit-unified-7d-reset", "1771549200")
+
+	result := calculateAnthropic429ResetTime(headers)
+	assertAnthropicResult(t, result, 1770998400)
+
+	if result.fiveHourReset == nil || !result.fiveHourReset.Equal(time.Unix(1770998400, 0)) {
+		t.Errorf("expected fiveHourReset=1770998400, got %v", result.fiveHourReset)
+	}
+}
+
+func TestCalculateAnthropic429ResetTime_Only7dExceeded(t *testing.T) {
+	headers := http.Header{}
+	headers.Set("anthropic-ratelimit-unified-5h-utilization", "0.50")
+	headers.Set("anthropic-ratelimit-unified-5h-reset", "1770998400")
+	headers.Set("anthropic-ratelimit-unified-7d-utilization", "1.05")
+	headers.Set("anthropic-ratelimit-unified-7d-reset", "1771549200")
+
+	result := calculateAnthropic429ResetTime(headers)
+	assertAnthropicResult(t, result, 1771549200)
+
+	// fiveHourReset should still be populated for session window calculation
+	if result.fiveHourReset == nil || !result.fiveHourReset.Equal(time.Unix(1770998400, 0)) {
+		t.Errorf("expected fiveHourReset=1770998400, got %v", result.fiveHourReset)
+	}
+}
+
+func TestCalculateAnthropic429ResetTime_BothExceeded(t *testing.T) {
+	headers := http.Header{}
+	headers.Set("anthropic-ratelimit-unified-5h-utilization", "1.10")
+	headers.Set("anthropic-ratelimit-unified-5h-reset", "1770998400")
+	headers.Set("anthropic-ratelimit-unified-7d-utilization", "1.02")
+	headers.Set("anthropic-ratelimit-unified-7d-reset", "1771549200")
+
+	result := calculateAnthropic429ResetTime(headers)
+	assertAnthropicResult(t, result, 1771549200)
+}
+
+func TestCalculateAnthropic429ResetTime_NoPerWindowHeaders(t *testing.T) {
+	headers := http.Header{}
+	headers.Set("anthropic-ratelimit-unified-reset", "1771549200")
+
+	result := calculateAnthropic429ResetTime(headers)
+	if result != nil {
+		t.Errorf("expected nil result when no per-window headers, got resetAt=%v", result.resetAt)
+	}
+}
+
+func TestCalculateAnthropic429ResetTime_NoHeaders(t *testing.T) {
+	result := calculateAnthropic429ResetTime(http.Header{})
+	if result != nil {
+		t.Errorf("expected nil result for empty headers, got resetAt=%v", result.resetAt)
+	}
+}
+
+func TestCalculateAnthropic429ResetTime_SurpassedThreshold(t *testing.T) {
+	headers := http.Header{}
+	headers.Set("anthropic-ratelimit-unified-5h-surpassed-threshold", "true")
+	headers.Set("anthropic-ratelimit-unified-5h-reset", "1770998400")
+	headers.Set("anthropic-ratelimit-unified-7d-surpassed-threshold", "false")
+	headers.Set("anthropic-ratelimit-unified-7d-reset", "1771549200")
+
+	result := calculateAnthropic429ResetTime(headers)
+	assertAnthropicResult(t, result, 1770998400)
+}
+
+func TestCalculateAnthropic429ResetTime_UtilizationExactlyOne(t *testing.T) {
+	headers := http.Header{}
+	headers.Set("anthropic-ratelimit-unified-5h-utilization", "1.0")
+	headers.Set("anthropic-ratelimit-unified-5h-reset", "1770998400")
+	headers.Set("anthropic-ratelimit-unified-7d-utilization", "0.5")
+	headers.Set("anthropic-ratelimit-unified-7d-reset", "1771549200")
+
+	result := calculateAnthropic429ResetTime(headers)
+	assertAnthropicResult(t, result, 1770998400)
+}
+
+func TestCalculateAnthropic429ResetTime_NeitherExceeded_UsesShorter(t *testing.T) {
+	headers := http.Header{}
+	headers.Set("anthropic-ratelimit-unified-5h-utilization", "0.95")
+	headers.Set("anthropic-ratelimit-unified-5h-reset", "1770998400") // sooner
+	headers.Set("anthropic-ratelimit-unified-7d-utilization", "0.80")
+	headers.Set("anthropic-ratelimit-unified-7d-reset", "1771549200") // later
+
+	result := calculateAnthropic429ResetTime(headers)
+	assertAnthropicResult(t, result, 1770998400)
+}
+
+func TestCalculateAnthropic429ResetTime_Only5hResetHeader(t *testing.T) {
+	headers := http.Header{}
+	headers.Set("anthropic-ratelimit-unified-5h-utilization", "1.05")
+	headers.Set("anthropic-ratelimit-unified-5h-reset", "1770998400")
+
+	result := calculateAnthropic429ResetTime(headers)
+	assertAnthropicResult(t, result, 1770998400)
+}
+
+func TestCalculateAnthropic429ResetTime_Only7dResetHeader(t *testing.T) {
+	headers := http.Header{}
+	headers.Set("anthropic-ratelimit-unified-7d-utilization", "1.03")
+	headers.Set("anthropic-ratelimit-unified-7d-reset", "1771549200")
+
+	result := calculateAnthropic429ResetTime(headers)
+	assertAnthropicResult(t, result, 1771549200)
+
+	if result.fiveHourReset != nil {
+		t.Errorf("expected fiveHourReset=nil when no 5h headers, got %v", result.fiveHourReset)
+	}
+}
+
+func TestIsAnthropicWindowExceeded(t *testing.T) {
+	tests := []struct {
+		name     string
+		headers  http.Header
+		window   string
+		expected bool
+	}{
+		{
+			name:     "utilization above 1.0",
+			headers:  makeHeader("anthropic-ratelimit-unified-5h-utilization", "1.02"),
+			window:   "5h",
+			expected: true,
+		},
+		{
+			name:     "utilization exactly 1.0",
+			headers:  makeHeader("anthropic-ratelimit-unified-5h-utilization", "1.0"),
+			window:   "5h",
+			expected: true,
+		},
+		{
+			name:     "utilization below 1.0",
+			headers:  makeHeader("anthropic-ratelimit-unified-5h-utilization", "0.99"),
+			window:   "5h",
+			expected: false,
+		},
+		{
+			name:     "surpassed-threshold true",
+			headers:  makeHeader("anthropic-ratelimit-unified-7d-surpassed-threshold", "true"),
+			window:   "7d",
+			expected: true,
+		},
+		{
+			name:     "surpassed-threshold True (case insensitive)",
+			headers:  makeHeader("anthropic-ratelimit-unified-7d-surpassed-threshold", "True"),
+			window:   "7d",
+			expected: true,
+		},
+		{
+			name:     "surpassed-threshold false",
+			headers:  makeHeader("anthropic-ratelimit-unified-7d-surpassed-threshold", "false"),
+			window:   "7d",
+			expected: false,
+		},
+		{
+			name:     "no headers",
+			headers:  http.Header{},
+			window:   "5h",
+			expected: false,
+		},
+	}
+
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			got := isAnthropicWindowExceeded(tc.headers, tc.window)
+			if got != tc.expected {
+				t.Errorf("expected %v, got %v", tc.expected, got)
+			}
+		})
+	}
+}
+
+// assertAnthropicResult is a test helper that verifies the result is non-nil and
+// has the expected resetAt unix timestamp.
+func assertAnthropicResult(t *testing.T, result *anthropic429Result, wantUnix int64) {
+	t.Helper()
+	if result == nil {
+		t.Fatal("expected non-nil result")
+		return // unreachable, but satisfies staticcheck SA5011
+	}
+	want := time.Unix(wantUnix, 0)
+	if !result.resetAt.Equal(want) {
+		t.Errorf("expected resetAt=%v, got %v", want, result.resetAt)
+	}
+}
+
+func makeHeader(key, value string) http.Header {
+	h := http.Header{}
+	h.Set(key, value)
+	return h
+}

backend/internal/service/usage_log.go

@@ -26,8 +26,8 @@ type UsageLog struct {
 	CacheCreationTokens int
 	CacheReadTokens     int
 
-	CacheCreation5mTokens int
+	CacheCreation5mTokens int `gorm:"column:cache_creation_5m_tokens"`
-	CacheCreation1hTokens int
+	CacheCreation1hTokens int `gorm:"column:cache_creation_1h_tokens"`
 
 	InputCost         float64
 	OutputCost        float64
@@ -46,6 +46,9 @@ type UsageLog struct {
 	UserAgent    *string
 	IPAddress    *string
 
+	// Cache TTL Override 标记（管理员强制替换了缓存 TTL 计费）
+	CacheTTLOverridden bool
+
 	// 图片生成字段
 	ImageCount int
 	ImageSize  *string

backend/migrations/054_drop_legacy_cache_columns.sql

@@ -0,0 +1,14 @@
+-- Drop legacy cache token columns that lack the underscore separator.
+-- These were created by GORM's automatic snake_case conversion:
+--   CacheCreation5mTokens → cache_creation5m_tokens  (incorrect)
+--   CacheCreation1hTokens → cache_creation1h_tokens  (incorrect)
+--
+-- The canonical columns are:
+--   cache_creation_5m_tokens  (defined in 001_init.sql)
+--   cache_creation_1h_tokens  (defined in 001_init.sql)
+--
+-- Migration 009 already copied data from legacy → canonical columns.
+-- This migration drops the legacy columns to avoid confusion.
+
+ALTER TABLE usage_logs DROP COLUMN IF EXISTS cache_creation5m_tokens;
+ALTER TABLE usage_logs DROP COLUMN IF EXISTS cache_creation1h_tokens;

backend/migrations/055_add_cache_ttl_overridden.sql

@@ -0,0 +1,2 @@
+-- Add cache_ttl_overridden flag to usage_logs for tracking cache TTL override per account.
+ALTER TABLE usage_logs ADD COLUMN IF NOT EXISTS cache_ttl_overridden BOOLEAN NOT NULL DEFAULT FALSE;

deploy/docker-compose.yml

@@ -158,6 +158,7 @@ services:
       - POSTGRES_USER=${POSTGRES_USER:-sub2api}
       - POSTGRES_PASSWORD=${POSTGRES_PASSWORD:?POSTGRES_PASSWORD is required}
       - POSTGRES_DB=${POSTGRES_DB:-sub2api}
+      - PGDATA=/var/lib/postgresql/data
       - TZ=${TZ:-Asia/Shanghai}
     networks:
       - sub2api-network

frontend/src/api/admin/accounts.ts

@@ -32,6 +32,7 @@ export async function list(
     platform?: string
     type?: string
     status?: string
+    group?: string
     search?: string
   },
   options?: {

frontend/src/api/admin/antigravity.ts

@@ -53,4 +53,18 @@ export async function exchangeCode(
   return data
 }
 
-export default { generateAuthUrl, exchangeCode }
+export async function refreshAntigravityToken(
+  refreshToken: string,
+  proxyId?: number | null
+): Promise<AntigravityTokenInfo> {
+  const payload: Record<string, any> = { refresh_token: refreshToken }
+  if (proxyId) payload.proxy_id = proxyId
+  
+  const { data } = await apiClient.post<AntigravityTokenInfo>(
+    '/admin/antigravity/oauth/refresh-token',
+    payload
+  )
+  return data
+}
+
+export default { generateAuthUrl, exchangeCode, refreshAntigravityToken }

frontend/src/components/account/AccountGroupsCell.vue

@@ -41,7 +41,7 @@
         >
           <div class="mb-2 flex items-center justify-between">
             <span class="text-xs font-medium text-gray-500 dark:text-gray-400">
-              {{ t('admin.accounts.allGroups', { count: groups.length }) }}
+              {{ t('admin.accounts.groupCountTotal', { count: groups.length }) }}
             </span>
             <button
               @click="showPopover = false"

frontend/src/components/account/BulkEditAccountModal.vue

@@ -708,6 +708,7 @@ const groupIds = ref<number[]>([])
 // All models list (combined Anthropic + OpenAI)
 const allModels = [
   { value: 'claude-opus-4-6', label: 'Claude Opus 4.6' },
+  { value: 'claude-sonnet-4-6', label: 'Claude Sonnet 4.6' },
   { value: 'claude-opus-4-5-20251101', label: 'Claude Opus 4.5' },
   { value: 'claude-sonnet-4-20250514', label: 'Claude Sonnet 4' },
   { value: 'claude-sonnet-4-5-20250929', label: 'Claude Sonnet 4.5' },
@@ -754,6 +755,13 @@ const presetMappings = [
     color:
       'bg-purple-100 text-purple-700 hover:bg-purple-200 dark:bg-purple-900/30 dark:text-purple-400'
   },
+  {
+    label: 'Sonnet 4.6',
+    from: 'claude-sonnet-4-6',
+    to: 'claude-sonnet-4-6',
+    color:
+      'bg-purple-100 text-purple-700 hover:bg-purple-200 dark:bg-purple-900/30 dark:text-purple-400'
+  },
   {
     label: 'Opus->Sonnet',
     from: 'claude-opus-4-5-20251101',

frontend/src/components/account/CreateAccountModal.vue

@@ -1527,6 +1527,46 @@
             </button>
           </div>
         </div>
+
+        <!-- Cache TTL Override -->
+        <div class="rounded-lg border border-gray-200 p-4 dark:border-dark-600">
+          <div class="flex items-center justify-between">
+            <div>
+              <label class="input-label mb-0">{{ t('admin.accounts.quotaControl.cacheTTLOverride.label') }}</label>
+              <p class="mt-1 text-xs text-gray-500 dark:text-gray-400">
+                {{ t('admin.accounts.quotaControl.cacheTTLOverride.hint') }}
+              </p>
+            </div>
+            <button
+              type="button"
+              @click="cacheTTLOverrideEnabled = !cacheTTLOverrideEnabled"
+              :class="[
+                'relative inline-flex h-6 w-11 flex-shrink-0 cursor-pointer rounded-full border-2 border-transparent transition-colors duration-200 ease-in-out focus:outline-none focus:ring-2 focus:ring-primary-500 focus:ring-offset-2',
+                cacheTTLOverrideEnabled ? 'bg-primary-600' : 'bg-gray-200 dark:bg-dark-600'
+              ]"
+            >
+              <span
+                :class="[
+                  'pointer-events-none inline-block h-5 w-5 transform rounded-full bg-white shadow ring-0 transition duration-200 ease-in-out',
+                  cacheTTLOverrideEnabled ? 'translate-x-5' : 'translate-x-0'
+                ]"
+              />
+            </button>
+          </div>
+          <div v-if="cacheTTLOverrideEnabled" class="mt-3">
+            <label class="input-label text-xs">{{ t('admin.accounts.quotaControl.cacheTTLOverride.target') }}</label>
+            <select
+              v-model="cacheTTLOverrideTarget"
+              class="mt-1 block w-full rounded-md border border-gray-300 bg-white px-3 py-2 text-sm shadow-sm focus:border-primary-500 focus:outline-none focus:ring-1 focus:ring-primary-500 dark:border-dark-500 dark:bg-dark-700 dark:text-white"
+            >
+              <option value="5m">5m</option>
+              <option value="1h">1h</option>
+            </select>
+            <p class="mt-1 text-xs text-gray-500 dark:text-gray-400">
+              {{ t('admin.accounts.quotaControl.cacheTTLOverride.targetHint') }}
+            </p>
+          </div>
+        </div>
       </div>
 
       <div>
@@ -1647,12 +1687,12 @@
         :show-proxy-warning="form.platform !== 'openai' && !!form.proxy_id"
         :allow-multiple="form.platform === 'anthropic'"
         :show-cookie-option="form.platform === 'anthropic'"
-        :show-refresh-token-option="form.platform === 'openai'"
+        :show-refresh-token-option="form.platform === 'openai' || form.platform === 'antigravity'"
         :platform="form.platform"
         :show-project-id="geminiOAuthType === 'code_assist'"
         @generate-url="handleGenerateUrl"
         @cookie-auth="handleCookieAuth"
-        @validate-refresh-token="handleOpenAIValidateRT"
+        @validate-refresh-token="handleValidateRefreshToken"
       />
 
     </div>
@@ -2146,6 +2186,8 @@ const maxSessions = ref<number | null>(null)
 const sessionIdleTimeout = ref<number | null>(null)
 const tlsFingerprintEnabled = ref(false)
 const sessionIdMaskingEnabled = ref(false)
+const cacheTTLOverrideEnabled = ref(false)
+const cacheTTLOverrideTarget = ref<string>('5m')
 
 // Gemini tier selection (used as fallback when auto-detection is unavailable/fails)
 const geminiTierGoogleOne = ref<'google_one_free' | 'google_ai_pro' | 'google_ai_ultra'>('google_one_free')
@@ -2597,6 +2639,8 @@ const resetForm = () => {
   sessionIdleTimeout.value = null
   tlsFingerprintEnabled.value = false
   sessionIdMaskingEnabled.value = false
+  cacheTTLOverrideEnabled.value = false
+  cacheTTLOverrideTarget.value = '5m'
   antigravityAccountType.value = 'oauth'
   upstreamBaseUrl.value = ''
   upstreamApiKey.value = ''
@@ -2802,6 +2846,14 @@ const handleGenerateUrl = async () => {
   }
 }
 
+const handleValidateRefreshToken = (rt: string) => {
+  if (form.platform === 'openai') {
+    handleOpenAIValidateRT(rt)
+  } else if (form.platform === 'antigravity') {
+    handleAntigravityValidateRT(rt)
+  }
+}
+
 const formatDateTimeLocal = formatDateTimeLocalInput
 const parseDateTimeLocal = parseDateTimeLocalInput
 
@@ -2950,6 +3002,95 @@ const handleOpenAIValidateRT = async (refreshTokenInput: string) => {
   }
 }
 
+// Antigravity 手动 RT 批量验证和创建
+const handleAntigravityValidateRT = async (refreshTokenInput: string) => {
+  if (!refreshTokenInput.trim()) return
+
+  // Parse multiple refresh tokens (one per line)
+  const refreshTokens = refreshTokenInput
+    .split('\n')
+    .map((rt) => rt.trim())
+    .filter((rt) => rt)
+
+  if (refreshTokens.length === 0) {
+    antigravityOAuth.error.value = t('admin.accounts.oauth.antigravity.pleaseEnterRefreshToken')
+    return
+  }
+
+  antigravityOAuth.loading.value = true
+  antigravityOAuth.error.value = ''
+
+  let successCount = 0
+  let failedCount = 0
+  const errors: string[] = []
+
+  try {
+    for (let i = 0; i < refreshTokens.length; i++) {
+      try {
+        const tokenInfo = await antigravityOAuth.validateRefreshToken(
+          refreshTokens[i],
+          form.proxy_id
+        )
+        if (!tokenInfo) {
+          failedCount++
+          errors.push(`#${i + 1}: ${antigravityOAuth.error.value || 'Validation failed'}`)
+          antigravityOAuth.error.value = ''
+          continue
+        }
+
+        const credentials = antigravityOAuth.buildCredentials(tokenInfo)
+        
+        // Generate account name with index for batch
+        const accountName = refreshTokens.length > 1 ? `${form.name} #${i + 1}` : form.name
+
+        // Note: Antigravity doesn't have buildExtraInfo, so we pass empty extra or rely on credentials
+        await adminAPI.accounts.create({
+          name: accountName,
+          notes: form.notes,
+          platform: 'antigravity',
+          type: 'oauth',
+          credentials,
+          extra: {},
+          proxy_id: form.proxy_id,
+          concurrency: form.concurrency,
+          priority: form.priority,
+          rate_multiplier: form.rate_multiplier,
+          group_ids: form.group_ids,
+          expires_at: form.expires_at,
+          auto_pause_on_expired: autoPauseOnExpired.value
+        })
+        successCount++
+      } catch (error: any) {
+        failedCount++
+        const errMsg = error.response?.data?.detail || error.message || 'Unknown error'
+        errors.push(`#${i + 1}: ${errMsg}`)
+      }
+    }
+
+    // Show results
+    if (successCount > 0 && failedCount === 0) {
+      appStore.showSuccess(
+        refreshTokens.length > 1
+          ? t('admin.accounts.oauth.batchSuccess', { count: successCount })
+          : t('admin.accounts.accountCreated')
+      )
+      emit('created')
+      handleClose()
+    } else if (successCount > 0 && failedCount > 0) {
+      appStore.showWarning(
+        t('admin.accounts.oauth.batchPartialSuccess', { success: successCount, failed: failedCount })
+      )
+      antigravityOAuth.error.value = errors.join('\n')
+      emit('created')
+    } else {
+      antigravityOAuth.error.value = errors.join('\n')
+      appStore.showError(t('admin.accounts.oauth.batchFailed'))
+    }
+  } finally {
+    antigravityOAuth.loading.value = false
+  }
+}
+
 // Gemini OAuth 授权码兑换
 const handleGeminiExchange = async (authCode: string) => {
   if (!authCode.trim() || !geminiOAuth.sessionId.value) return
@@ -3077,6 +3218,12 @@ const handleAnthropicExchange = async (authCode: string) => {
       extra.session_id_masking_enabled = true
     }
 
+    // Add cache TTL override settings
+    if (cacheTTLOverrideEnabled.value) {
+      extra.cache_ttl_override_enabled = true
+      extra.cache_ttl_override_target = cacheTTLOverrideTarget.value
+    }
+
     const credentials = {
       ...tokenInfo,
       ...(interceptWarmupRequests.value ? { intercept_warmup_requests: true } : {})
@@ -3170,6 +3317,12 @@ const handleCookieAuth = async (sessionKey: string) => {
           extra.session_id_masking_enabled = true
         }
 
+        // Add cache TTL override settings
+        if (cacheTTLOverrideEnabled.value) {
+          extra.cache_ttl_override_enabled = true
+          extra.cache_ttl_override_target = cacheTTLOverrideTarget.value
+        }
+
         const accountName = keys.length > 1 ? `${form.name} #${i + 1}` : form.name
 
         // Merge interceptWarmupRequests into credentials

frontend/src/components/account/EditAccountModal.vue

@@ -904,6 +904,46 @@
             </button>
           </div>
         </div>
+
+        <!-- Cache TTL Override -->
+        <div class="rounded-lg border border-gray-200 p-4 dark:border-dark-600">
+          <div class="flex items-center justify-between">
+            <div>
+              <label class="input-label mb-0">{{ t('admin.accounts.quotaControl.cacheTTLOverride.label') }}</label>
+              <p class="mt-1 text-xs text-gray-500 dark:text-gray-400">
+                {{ t('admin.accounts.quotaControl.cacheTTLOverride.hint') }}
+              </p>
+            </div>
+            <button
+              type="button"
+              @click="cacheTTLOverrideEnabled = !cacheTTLOverrideEnabled"
+              :class="[
+                'relative inline-flex h-6 w-11 flex-shrink-0 cursor-pointer rounded-full border-2 border-transparent transition-colors duration-200 ease-in-out focus:outline-none focus:ring-2 focus:ring-primary-500 focus:ring-offset-2',
+                cacheTTLOverrideEnabled ? 'bg-primary-600' : 'bg-gray-200 dark:bg-dark-600'
+              ]"
+            >
+              <span
+                :class="[
+                  'pointer-events-none inline-block h-5 w-5 transform rounded-full bg-white shadow ring-0 transition duration-200 ease-in-out',
+                  cacheTTLOverrideEnabled ? 'translate-x-5' : 'translate-x-0'
+                ]"
+              />
+            </button>
+          </div>
+          <div v-if="cacheTTLOverrideEnabled" class="mt-3">
+            <label class="input-label text-xs">{{ t('admin.accounts.quotaControl.cacheTTLOverride.target') }}</label>
+            <select
+              v-model="cacheTTLOverrideTarget"
+              class="mt-1 block w-full rounded-md border border-gray-300 bg-white px-3 py-2 text-sm shadow-sm focus:border-primary-500 focus:outline-none focus:ring-1 focus:ring-primary-500 dark:border-dark-500 dark:bg-dark-700 dark:text-white"
+            >
+              <option value="5m">5m</option>
+              <option value="1h">1h</option>
+            </select>
+            <p class="mt-1 text-xs text-gray-500 dark:text-gray-400">
+              {{ t('admin.accounts.quotaControl.cacheTTLOverride.targetHint') }}
+            </p>
+          </div>
+        </div>
       </div>
 
       <div class="border-t border-gray-200 pt-4 dark:border-dark-600">
@@ -1102,6 +1142,8 @@ const maxSessions = ref<number | null>(null)
 const sessionIdleTimeout = ref<number | null>(null)
 const tlsFingerprintEnabled = ref(false)
 const sessionIdMaskingEnabled = ref(false)
+const cacheTTLOverrideEnabled = ref(false)
+const cacheTTLOverrideTarget = ref<string>('5m')
 
 // Computed: current preset mappings based on platform
 const presetMappings = computed(() => getPresetMappingsByPlatform(props.account?.platform || 'anthropic'))
@@ -1489,6 +1531,8 @@ function loadQuotaControlSettings(account: Account) {
   sessionIdleTimeout.value = null
   tlsFingerprintEnabled.value = false
   sessionIdMaskingEnabled.value = false
+  cacheTTLOverrideEnabled.value = false
+  cacheTTLOverrideTarget.value = '5m'
 
   // Only applies to Anthropic OAuth/SetupToken accounts
   if (account.platform !== 'anthropic' || (account.type !== 'oauth' && account.type !== 'setup-token')) {
@@ -1517,6 +1561,12 @@ function loadQuotaControlSettings(account: Account) {
   if (account.session_id_masking_enabled === true) {
     sessionIdMaskingEnabled.value = true
   }
+
+  // Load cache TTL override setting
+  if (account.cache_ttl_override_enabled === true) {
+    cacheTTLOverrideEnabled.value = true
+    cacheTTLOverrideTarget.value = account.cache_ttl_override_target || '5m'
+  }
 }
 
 function formatTempUnschedKeywords(value: unknown) {
@@ -1723,6 +1773,15 @@ const handleSubmit = async () => {
         delete newExtra.session_id_masking_enabled
       }
 
+      // Cache TTL override setting
+      if (cacheTTLOverrideEnabled.value) {
+        newExtra.cache_ttl_override_enabled = true
+        newExtra.cache_ttl_override_target = cacheTTLOverrideTarget.value
+      } else {
+        delete newExtra.cache_ttl_override_enabled
+        delete newExtra.cache_ttl_override_target
+      }
+
       updatePayload.extra = newExtra
     }
 

frontend/src/components/account/OAuthAuthorizationFlow.vue

@@ -45,19 +45,19 @@
                 class="text-blue-600 focus:ring-blue-500"
               />
               <span class="text-sm text-blue-900 dark:text-blue-200">{{
-                t('admin.accounts.oauth.openai.refreshTokenAuth')
+                t(getOAuthKey('refreshTokenAuth'))
               }}</span>
             </label>
           </div>
         </div>
 
-        <!-- Refresh Token Input (OpenAI only) -->
+        <!-- Refresh Token Input (OpenAI / Antigravity) -->
         <div v-if="inputMethod === 'refresh_token'" class="space-y-4">
           <div
             class="rounded-lg border border-blue-300 bg-white/80 p-4 dark:border-blue-600 dark:bg-gray-800/80"
           >
             <p class="mb-3 text-sm text-blue-700 dark:text-blue-300">
-              {{ t('admin.accounts.oauth.openai.refreshTokenDesc') }}
+              {{ t(getOAuthKey('refreshTokenDesc')) }}
             </p>
 
             <!-- Refresh Token Input -->
@@ -78,7 +78,7 @@
                 v-model="refreshTokenInput"
                 rows="3"
                 class="input w-full resize-y font-mono text-sm"
-                :placeholder="t('admin.accounts.oauth.openai.refreshTokenPlaceholder')"
+                :placeholder="t(getOAuthKey('refreshTokenPlaceholder'))"
               ></textarea>
               <p
                 v-if="parsedRefreshTokenCount > 1"
@@ -128,8 +128,8 @@
               <Icon v-else name="sparkles" size="sm" class="mr-2" />
               {{
                 loading
-                  ? t('admin.accounts.oauth.openai.validating')
+                  ? t(getOAuthKey('validating'))
-                  : t('admin.accounts.oauth.openai.validateAndCreate')
+                  : t(getOAuthKey('validateAndCreate'))
               }}
             </button>
           </div>

frontend/src/components/admin/account/AccountTableFilters.vue

@@ -10,16 +10,21 @@
     <Select :model-value="filters.platform" class="w-40" :options="pOpts" @update:model-value="updatePlatform" @change="$emit('change')" />
     <Select :model-value="filters.type" class="w-40" :options="tOpts" @update:model-value="updateType" @change="$emit('change')" />
     <Select :model-value="filters.status" class="w-40" :options="sOpts" @update:model-value="updateStatus" @change="$emit('change')" />
+    <Select :model-value="filters.group" class="w-40" :options="gOpts" @update:model-value="updateGroup" @change="$emit('change')" />
   </div>
 </template>
 
 <script setup lang="ts">
 import { computed } from 'vue'; import { useI18n } from 'vue-i18n'; import Select from '@/components/common/Select.vue'; import SearchInput from '@/components/common/SearchInput.vue'
-const props = defineProps(['searchQuery', 'filters']); const emit = defineEmits(['update:searchQuery', 'update:filters', 'change']); const { t } = useI18n()
+import type { AdminGroup } from '@/types'
+const props = defineProps<{ searchQuery: string; filters: Record<string, any>; groups?: AdminGroup[] }>()
+const emit = defineEmits(['update:searchQuery', 'update:filters', 'change']); const { t } = useI18n()
 const updatePlatform = (value: string | number | boolean | null) => { emit('update:filters', { ...props.filters, platform: value }) }
 const updateType = (value: string | number | boolean | null) => { emit('update:filters', { ...props.filters, type: value }) }
 const updateStatus = (value: string | number | boolean | null) => { emit('update:filters', { ...props.filters, status: value }) }
+const updateGroup = (value: string | number | boolean | null) => { emit('update:filters', { ...props.filters, group: value }) }
 const pOpts = computed(() => [{ value: '', label: t('admin.accounts.allPlatforms') }, { value: 'anthropic', label: 'Anthropic' }, { value: 'openai', label: 'OpenAI' }, { value: 'gemini', label: 'Gemini' }, { value: 'antigravity', label: 'Antigravity' }])
 const tOpts = computed(() => [{ value: '', label: t('admin.accounts.allTypes') }, { value: 'oauth', label: t('admin.accounts.oauthType') }, { value: 'setup-token', label: t('admin.accounts.setupToken') }, { value: 'apikey', label: t('admin.accounts.apiKey') }])
-const sOpts = computed(() => [{ value: '', label: t('admin.accounts.allStatus') }, { value: 'active', label: t('admin.accounts.status.active') }, { value: 'inactive', label: t('admin.accounts.status.inactive') }, { value: 'error', label: t('admin.accounts.status.error') }])
+const sOpts = computed(() => [{ value: '', label: t('admin.accounts.allStatus') }, { value: 'active', label: t('admin.accounts.status.active') }, { value: 'inactive', label: t('admin.accounts.status.inactive') }, { value: 'error', label: t('admin.accounts.status.error') }, { value: 'rate_limited', label: t('admin.accounts.status.rateLimited') }])
+const gOpts = computed(() => [{ value: '', label: t('admin.accounts.allGroups') }, ...(props.groups || []).map(g => ({ value: String(g.id), label: g.name }))])
 </script>

frontend/src/components/admin/usage/UsageTable.vue

@@ -70,6 +70,8 @@
                 <div v-if="row.cache_creation_tokens > 0" class="inline-flex items-center gap-1">
                   <svg class="h-3.5 w-3.5 text-amber-500" fill="none" stroke="currentColor" viewBox="0 0 24 24"><path stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M11 5H6a2 2 0 00-2 2v11a2 2 0 002 2h11a2 2 0 002-2v-5m-1.414-9.414a2 2 0 112.828 2.828L11.828 15H9v-2.828l8.586-8.586z" /></svg>
                   <span class="font-medium text-amber-600 dark:text-amber-400">{{ formatCacheTokens(row.cache_creation_tokens) }}</span>
+                  <span v-if="row.cache_creation_1h_tokens > 0" class="inline-flex items-center rounded px-1 py-px text-[10px] font-medium leading-tight bg-orange-100 text-orange-600 ring-1 ring-inset ring-orange-200 dark:bg-orange-500/20 dark:text-orange-400 dark:ring-orange-500/30">1h</span>
+                  <span v-if="row.cache_ttl_overridden" :title="t('usage.cacheTtlOverriddenHint')" class="inline-flex items-center rounded px-1 py-px text-[10px] font-medium leading-tight bg-rose-100 text-rose-600 ring-1 ring-inset ring-rose-200 dark:bg-rose-500/20 dark:text-rose-400 dark:ring-rose-500/30 cursor-help">R</span>
                 </div>
               </div>
             </div>
@@ -157,9 +159,36 @@
               <span class="text-gray-400">{{ t('admin.usage.outputTokens') }}</span>
               <span class="font-medium text-white">{{ tokenTooltipData.output_tokens.toLocaleString() }}</span>
             </div>
-            <div v-if="tokenTooltipData && tokenTooltipData.cache_creation_tokens > 0" class="flex items-center justify-between gap-4">
+            <div v-if="tokenTooltipData && tokenTooltipData.cache_creation_tokens > 0">
-              <span class="text-gray-400">{{ t('admin.usage.cacheCreationTokens') }}</span>
+              <!-- 有 5m/1h 明细时，展开显示 -->
-              <span class="font-medium text-white">{{ tokenTooltipData.cache_creation_tokens.toLocaleString() }}</span>
+              <template v-if="tokenTooltipData.cache_creation_5m_tokens > 0 || tokenTooltipData.cache_creation_1h_tokens > 0">
+                <div v-if="tokenTooltipData.cache_creation_5m_tokens > 0" class="flex items-center justify-between gap-4">
+                  <span class="text-gray-400 flex items-center gap-1.5">
+                    {{ t('admin.usage.cacheCreation5mTokens') }}
+                    <span class="inline-flex items-center rounded px-1 py-px text-[10px] font-medium leading-tight bg-amber-500/20 text-amber-400 ring-1 ring-inset ring-amber-500/30">5m</span>
+                  </span>
+                  <span class="font-medium text-white">{{ tokenTooltipData.cache_creation_5m_tokens.toLocaleString() }}</span>
+                </div>
+                <div v-if="tokenTooltipData.cache_creation_1h_tokens > 0" class="flex items-center justify-between gap-4">
+                  <span class="text-gray-400 flex items-center gap-1.5">
+                    {{ t('admin.usage.cacheCreation1hTokens') }}
+                    <span class="inline-flex items-center rounded px-1 py-px text-[10px] font-medium leading-tight bg-orange-500/20 text-orange-400 ring-1 ring-inset ring-orange-500/30">1h</span>
+                  </span>
+                  <span class="font-medium text-white">{{ tokenTooltipData.cache_creation_1h_tokens.toLocaleString() }}</span>
+                </div>
+              </template>
+              <!-- 无明细时，只显示聚合值 -->
+              <div v-else class="flex items-center justify-between gap-4">
+                <span class="text-gray-400">{{ t('admin.usage.cacheCreationTokens') }}</span>
+                <span class="font-medium text-white">{{ tokenTooltipData.cache_creation_tokens.toLocaleString() }}</span>
+              </div>
+            </div>
+            <div v-if="tokenTooltipData && tokenTooltipData.cache_ttl_overridden" class="flex items-center justify-between gap-4">
+              <span class="text-gray-400 flex items-center gap-1.5">
+                {{ t('usage.cacheTtlOverriddenLabel') }}
+                <span class="inline-flex items-center rounded px-1 py-px text-[10px] font-medium leading-tight bg-rose-500/20 text-rose-400 ring-1 ring-inset ring-rose-500/30">R-{{ tokenTooltipData.cache_creation_1h_tokens > 0 ? '5m' : '1H' }}</span>
+              </span>
+              <span class="font-medium text-rose-400">{{ tokenTooltipData.cache_creation_1h_tokens > 0 ? t('usage.cacheTtlOverridden1h') : t('usage.cacheTtlOverridden5m') }}</span>
             </div>
             <div v-if="tokenTooltipData && tokenTooltipData.cache_read_tokens > 0" class="flex items-center justify-between gap-4">
               <span class="text-gray-400">{{ t('admin.usage.cacheReadTokens') }}</span>

frontend/src/components/common/StatCard.vue

@@ -6,7 +6,7 @@
     <div class="min-w-0 flex-1">
       <p class="stat-label truncate">{{ title }}</p>
       <div class="mt-1 flex items-baseline gap-2">
-        <p class="stat-value">{{ formattedValue }}</p>
+        <p class="stat-value" :title="String(formattedValue)">{{ formattedValue }}</p>
         <span v-if="change !== undefined" :class="['stat-trend', trendClass]">
           <Icon
             v-if="changeType !== 'neutral'"

frontend/src/components/layout/AppSidebar.vue

@@ -10,7 +10,7 @@
     <div class="sidebar-header">
       <!-- Custom Logo or Default Logo -->
       <div class="flex h-9 w-9 items-center justify-center overflow-hidden rounded-xl shadow-glow">
-        <img :src="siteLogo || '/logo.png'" alt="Logo" class="h-full w-full object-contain" />
+        <img v-if="settingsLoaded" :src="siteLogo || '/logo.png'" alt="Logo" class="h-full w-full object-contain" />
       </div>
       <transition name="fade">
         <div v-if="!sidebarCollapsed" class="flex flex-col">
@@ -167,6 +167,7 @@ const isDark = ref(document.documentElement.classList.contains('dark'))
 const siteName = computed(() => appStore.siteName)
 const siteLogo = computed(() => appStore.siteLogo)
 const siteVersion = computed(() => appStore.siteVersion)
+const settingsLoaded = computed(() => appStore.publicSettingsLoaded)
 
 // SVG Icon Components
 const DashboardIcon = {

frontend/src/components/layout/AuthLayout.vue

@@ -29,17 +29,19 @@
       <!-- Logo/Brand -->
       <div class="mb-8 text-center">
         <!-- Custom Logo or Default Logo -->
-        <div
+        <template v-if="settingsLoaded">
-          class="mb-4 inline-flex h-16 w-16 items-center justify-center overflow-hidden rounded-2xl shadow-lg shadow-primary-500/30"
+          <div
-        >
+            class="mb-4 inline-flex h-16 w-16 items-center justify-center overflow-hidden rounded-2xl shadow-lg shadow-primary-500/30"
-          <img :src="siteLogo || '/logo.png'" alt="Logo" class="h-full w-full object-contain" />
+          >
-        </div>
+            <img :src="siteLogo || '/logo.png'" alt="Logo" class="h-full w-full object-contain" />
-        <h1 class="text-gradient mb-2 text-3xl font-bold">
+          </div>
-          {{ siteName }}
+          <h1 class="text-gradient mb-2 text-3xl font-bold">
-        </h1>
+            {{ siteName }}
-        <p class="text-sm text-gray-500 dark:text-dark-400">
+          </h1>
-          {{ siteSubtitle }}
+          <p class="text-sm text-gray-500 dark:text-dark-400">
-        </p>
+            {{ siteSubtitle }}
+          </p>
+        </template>
       </div>
 
       <!-- Card Container -->
@@ -61,25 +63,21 @@
 </template>
 
 <script setup lang="ts">
-import { ref, computed, onMounted } from 'vue'
+import { computed, onMounted } from 'vue'
-import { getPublicSettings } from '@/api/auth'
+import { useAppStore } from '@/stores'
 import { sanitizeUrl } from '@/utils/url'
 
-const siteName = ref('Sub2API')
+const appStore = useAppStore()
-const siteLogo = ref('')
+
-const siteSubtitle = ref('Subscription to API Conversion Platform')
+const siteName = computed(() => appStore.siteName || 'Sub2API')
+const siteLogo = computed(() => sanitizeUrl(appStore.siteLogo || '', { allowRelative: true, allowDataUrl: true }))
+const siteSubtitle = computed(() => appStore.cachedPublicSettings?.site_subtitle || 'Subscription to API Conversion Platform')
+const settingsLoaded = computed(() => appStore.publicSettingsLoaded)
 
 const currentYear = computed(() => new Date().getFullYear())
 
-onMounted(async () => {
+onMounted(() => {
-  try {
+  appStore.fetchPublicSettings()
-    const settings = await getPublicSettings()
-    siteName.value = settings.site_name || 'Sub2API'
-    siteLogo.value = sanitizeUrl(settings.site_logo || '', { allowRelative: true })
-    siteSubtitle.value = settings.site_subtitle || 'Subscription to API Conversion Platform'
-  } catch (error) {
-    console.error('Failed to load public settings:', error)
-  }
 })
 </script>
 

frontend/src/composables/useAntigravityOAuth.ts

@@ -83,6 +83,35 @@ export function useAntigravityOAuth() {
     }
   }
 
+  const validateRefreshToken = async (
+    refreshToken: string,
+    proxyId?: number | null
+  ): Promise<AntigravityTokenInfo | null> => {
+    if (!refreshToken.trim()) {
+      error.value = t('admin.accounts.oauth.antigravity.pleaseEnterRefreshToken')
+      return null
+    }
+
+    loading.value = true
+    error.value = ''
+
+    try {
+      const tokenInfo = await adminAPI.antigravity.refreshAntigravityToken(
+        refreshToken.trim(),
+        proxyId
+      )
+      return tokenInfo as AntigravityTokenInfo
+    } catch (err: any) {
+      error.value =
+        err.response?.data?.detail || t('admin.accounts.oauth.antigravity.failedToValidateRT')
+      // Don't show global error toast for batch validation to avoid spamming
+      // appStore.showError(error.value)
+      return null
+    } finally {
+      loading.value = false
+    }
+  }
+
   const buildCredentials = (tokenInfo: AntigravityTokenInfo): Record<string, unknown> => {
     let expiresAt: string | undefined
     if (typeof tokenInfo.expires_at === 'number' && Number.isFinite(tokenInfo.expires_at)) {
@@ -110,6 +139,7 @@ export function useAntigravityOAuth() {
     resetState,
     generateAuthUrl,
     exchangeAuthCode,
+    validateRefreshToken,
     buildCredentials
   }
 }

frontend/src/composables/useModelWhitelist.ts

@@ -39,6 +39,7 @@ export const claudeModels = [
   'claude-sonnet-4-5-20250929', 'claude-haiku-4-5-20251001',
   'claude-opus-4-5-20251101',
   'claude-opus-4-6',
+  'claude-sonnet-4-6',
   'claude-2.1', 'claude-2.0', 'claude-instant-1.2'
 ]
 
@@ -233,6 +234,7 @@ export const allModels = allModelsList.map(m => ({ value: m, label: m }))
 const anthropicPresetMappings = [
   { label: 'Sonnet 4', from: 'claude-sonnet-4-20250514', to: 'claude-sonnet-4-20250514', color: 'bg-blue-100 text-blue-700 hover:bg-blue-200 dark:bg-blue-900/30 dark:text-blue-400' },
   { label: 'Sonnet 4.5', from: 'claude-sonnet-4-5-20250929', to: 'claude-sonnet-4-5-20250929', color: 'bg-indigo-100 text-indigo-700 hover:bg-indigo-200 dark:bg-indigo-900/30 dark:text-indigo-400' },
+  { label: 'Sonnet 4.6', from: 'claude-sonnet-4-6', to: 'claude-sonnet-4-6', color: 'bg-indigo-100 text-indigo-700 hover:bg-indigo-200 dark:bg-indigo-900/30 dark:text-indigo-400' },
   { label: 'Opus 4.5', from: 'claude-opus-4-5-20251101', to: 'claude-opus-4-5-20251101', color: 'bg-purple-100 text-purple-700 hover:bg-purple-200 dark:bg-purple-900/30 dark:text-purple-400' },
   { label: 'Opus 4.6', from: 'claude-opus-4-6', to: 'claude-opus-4-6', color: 'bg-purple-100 text-purple-700 hover:bg-purple-200 dark:bg-purple-900/30 dark:text-purple-400' },
   { label: 'Haiku 3.5', from: 'claude-3-5-haiku-20241022', to: 'claude-3-5-haiku-20241022', color: 'bg-green-100 text-green-700 hover:bg-green-200 dark:bg-green-900/30 dark:text-green-400' },

frontend/src/i18n/locales/en.ts

@@ -576,6 +576,10 @@ export default {
     description: 'View and analyze your API usage history',
     costDetails: 'Cost Breakdown',
     tokenDetails: 'Token Breakdown',
+    cacheTtlOverriddenHint: 'Cache TTL Override enabled',
+    cacheTtlOverriddenLabel: 'TTL Override',
+    cacheTtlOverridden5m: 'Billed as 5m',
+    cacheTtlOverridden1h: 'Billed as 1h',
     totalRequests: 'Total Requests',
     totalTokens: 'Total Tokens',
     totalCost: 'Total Cost',
@@ -841,7 +845,7 @@ export default {
       createUser: 'Create User',
       editUser: 'Edit User',
       deleteUser: 'Delete User',
-      searchUsers: 'Search users...',
+      searchUsers: 'Search by email, username, notes, or API key...',
       allRoles: 'All Roles',
       allStatus: 'All Status',
       admin: 'Admin',
@@ -1335,6 +1339,7 @@ export default {
       allPlatforms: 'All Platforms',
       allTypes: 'All Types',
       allStatus: 'All Status',
+      allGroups: 'All Groups',
       oauthType: 'OAuth',
       setupToken: 'Setup Token',
       apiKey: 'API Key',
@@ -1344,7 +1349,7 @@ export default {
       schedulableEnabled: 'Scheduling enabled',
       schedulableDisabled: 'Scheduling disabled',
       failedToToggleSchedulable: 'Failed to toggle scheduling status',
-      allGroups: '{count} groups total',
+      groupCountTotal: '{count} groups total',
       platforms: {
         anthropic: 'Anthropic',
         claude: 'Claude',
@@ -1594,6 +1599,12 @@ export default {
         sessionIdMasking: {
           label: 'Session ID Masking',
           hint: 'When enabled, fixes the session ID in metadata.user_id for 15 minutes, making upstream think requests come from the same session'
+        },
+        cacheTTLOverride: {
+          label: 'Cache TTL Override',
+          hint: 'Force all cache creation tokens to be billed as the selected TTL tier (5m or 1h)',
+          target: 'Target TTL',
+          targetHint: 'Select the TTL tier for billing'
         }
       },
       expired: 'Expired',
@@ -1774,13 +1785,20 @@ export default {
           authCode: 'Authorization URL or Code',
           authCodePlaceholder:
             'Option 1: Copy the complete URL\n(http://localhost:xxx/auth/callback?code=...)\nOption 2: Copy only the code parameter value',
-          authCodeHint: 'You can copy the entire URL or just the code parameter value, the system will auto-detect',
+                    authCodeHint: 'You can copy the entire URL or just the code parameter value, the system will auto-detect',
-          failedToGenerateUrl: 'Failed to generate Antigravity auth URL',
+                    failedToGenerateUrl: 'Failed to generate Antigravity auth URL',
-          missingExchangeParams: 'Missing code, session ID, or state',
+                    missingExchangeParams: 'Missing code, session ID, or state',
-          failedToExchangeCode: 'Failed to exchange Antigravity auth code'
+                    failedToExchangeCode: 'Failed to exchange Antigravity auth code',
-        }
+                    // Refresh Token auth
-	      },
+                    refreshTokenAuth: 'Manual RT',
-      // Gemini specific (platform-wide)
+                    refreshTokenDesc: 'Enter your existing Antigravity Refresh Token. Supports batch input (one per line). The system will automatically validate and create accounts.',
+                    refreshTokenPlaceholder: 'Paste your Antigravity Refresh Token...\nSupports multiple tokens, one per line',
+                    validating: 'Validating...',
+                    validateAndCreate: 'Validate & Create',
+                    pleaseEnterRefreshToken: 'Please enter Refresh Token',
+                    failedToValidateRT: 'Failed to validate Refresh Token'
+                  }
+                },      // Gemini specific (platform-wide)
       gemini: {
         helpButton: 'Help',
         helpDialog: {
@@ -2129,7 +2147,7 @@ export default {
       title: 'Redeem Code Management',
       description: 'Generate and manage redeem codes',
       generateCodes: 'Generate Codes',
-      searchCodes: 'Search codes...',
+      searchCodes: 'Search codes or email...',
       allTypes: 'All Types',
       allStatus: 'All Status',
       balance: 'Balance',
@@ -2352,6 +2370,8 @@ export default {
       inputTokens: 'Input Tokens',
       outputTokens: 'Output Tokens',
       cacheCreationTokens: 'Cache Creation Tokens',
+      cacheCreation5mTokens: 'Cache Write',
+      cacheCreation1hTokens: 'Cache Write',
       cacheReadTokens: 'Cache Read Tokens',
       failedToLoad: 'Failed to load usage records',
       billingType: 'Billing Type',

frontend/src/i18n/locales/zh.ts

@@ -582,6 +582,10 @@ export default {
     description: '查看和分析您的 API 使用历史',
     costDetails: '成本明细',
     tokenDetails: 'Token 明细',
+    cacheTtlOverriddenHint: '缓存 TTL Override 已启用',
+    cacheTtlOverriddenLabel: 'TTL 替换',
+    cacheTtlOverridden5m: '按 5m 计费',
+    cacheTtlOverridden1h: '按 1h 计费',
     totalRequests: '总请求数',
     totalTokens: '总 Token',
     totalCost: '总消费',
@@ -865,8 +869,8 @@ export default {
       editUser: '编辑用户',
       deleteUser: '删除用户',
       deleteConfirmMessage: "确定要删除用户 '{email}' 吗？此操作无法撤销。",
-      searchPlaceholder: '搜索用户邮箱或用户名、备注、支持模糊查询...',
+      searchPlaceholder: '邮箱/用户名/备注/API Key 模糊搜索...',
-      searchUsers: '搜索用户邮箱或用户名、备注、支持模糊查询',
+      searchUsers: '邮箱/用户名/备注/API Key 模糊搜索',
       roleFilter: '角色筛选',
       allRoles: '全部角色',
       allStatus: '全部状态',
@@ -1426,6 +1430,7 @@ export default {
       allPlatforms: '全部平台',
       allTypes: '全部类型',
       allStatus: '全部状态',
+      allGroups: '全部分组',
       oauthType: 'OAuth',
       // Schedulable toggle
       schedulable: '参与调度',
@@ -1433,7 +1438,7 @@ export default {
       schedulableEnabled: '调度已开启',
       schedulableDisabled: '调度已关闭',
       failedToToggleSchedulable: '切换调度状态失败',
-      allGroups: '共 {count} 个分组',
+      groupCountTotal: '共 {count} 个分组',
       columns: {
         name: '名称',
         platformType: '平台/类型',
@@ -1740,6 +1745,12 @@ export default {
         sessionIdMasking: {
           label: '会话 ID 伪装',
           hint: '启用后将在 15 分钟内固定 metadata.user_id 中的 session ID，使上游认为请求来自同一会话'
+        },
+        cacheTTLOverride: {
+          label: '缓存 TTL 强制替换',
+          hint: '将所有缓存创建 token 强制按指定的 TTL 类型（5分钟或1小时）计费',
+          target: '目标 TTL',
+          targetHint: '选择计费使用的 TTL 类型'
         }
       },
       expired: '已过期',
@@ -1913,7 +1924,15 @@ export default {
           authCodeHint: '您可以直接复制整个链接或仅复制 code 参数值，系统会自动识别',
           failedToGenerateUrl: '生成 Antigravity 授权链接失败',
           missingExchangeParams: '缺少 code / session_id / state',
-          failedToExchangeCode: 'Antigravity 授权码兑换失败'
+          failedToExchangeCode: 'Antigravity 授权码兑换失败',
+          // Refresh Token auth
+          refreshTokenAuth: '手动输入 RT',
+          refreshTokenDesc: '输入您已有的 Antigravity Refresh Token，支持批量输入（每行一个），系统将自动验证并创建账号。',
+          refreshTokenPlaceholder: '粘贴您的 Antigravity Refresh Token...\n支持多个，每行一个',
+          validating: '验证中...',
+          validateAndCreate: '验证并创建账号',
+          pleaseEnterRefreshToken: '请输入 Refresh Token',
+          failedToValidateRT: '验证 Refresh Token 失败'
         }
       },
       // Gemini specific (platform-wide)
@@ -2292,7 +2311,7 @@ export default {
       allStatus: '全部状态',
       unused: '未使用',
       used: '已使用',
-      searchCodes: '搜索兑换码...',
+      searchCodes: '搜索兑换码或邮箱...',
       exportCsv: '导出 CSV',
       deleteAllUnused: '删除全部未使用',
       deleteCodeConfirm: '确定要删除此兑换码吗？此操作无法撤销。',
@@ -2518,6 +2537,8 @@ export default {
       inputTokens: '输入 Token',
       outputTokens: '输出 Token',
       cacheCreationTokens: '缓存创建 Token',
+      cacheCreation5mTokens: '缓存创建',
+      cacheCreation1hTokens: '缓存创建',
       cacheReadTokens: '缓存读取 Token',
       failedToLoad: '加载使用记录失败',
       billingType: '计费类型',

frontend/src/style.css

@@ -243,7 +243,7 @@
   }
 
   .stat-value {
-    @apply text-2xl font-bold text-gray-900 dark:text-white;
+    @apply text-2xl font-bold text-gray-900 dark:text-white truncate;
   }
 
   .stat-label {

frontend/src/types/index.ts

@@ -614,6 +614,10 @@ export interface Account {
   // 启用后将在15分钟内固定 metadata.user_id 中的 session ID
   session_id_masking_enabled?: boolean | null
 
+  // 缓存 TTL 强制替换（仅 Anthropic OAuth/SetupToken 账号有效）
+  cache_ttl_override_enabled?: boolean | null
+  cache_ttl_override_target?: string | null
+
   // 运行时状态（仅当启用对应限制时返回）
   current_window_cost?: number | null // 当前窗口费用
   active_sessions?: number | null // 当前活跃会话数
@@ -827,6 +831,9 @@ export interface UsageLog {
   // User-Agent
   user_agent: string | null
 
+  // Cache TTL Override
+  cache_ttl_overridden: boolean
+
   created_at: string
 
   user?: User

frontend/src/utils/url.ts

@@ -6,6 +6,7 @@
  */
 type SanitizeOptions = {
   allowRelative?: boolean
+  allowDataUrl?: boolean
 }
 
 export function sanitizeUrl(value: string, options: SanitizeOptions = {}): string {
@@ -18,6 +19,11 @@ export function sanitizeUrl(value: string, options: SanitizeOptions = {}): strin
     return trimmed
   }
 
+  // 允许 data:image/ 开头的 data URL（仅限图片类型）
+  if (options.allowDataUrl && trimmed.startsWith('data:image/')) {
+    return trimmed
+  }
+
   // 只接受绝对 URL，不使用 base URL 来避免相对路径被解析为当前域名
   // 检查是否以 http:// 或 https:// 开头
   if (!trimmed.match(/^https?:\/\//i)) {

frontend/src/views/admin/AccountsView.vue

@@ -6,6 +6,7 @@
           <AccountTableFilters
             v-model:searchQuery="params.search"
             :filters="params"
+            :groups="groups"
             @update:filters="(newFilters) => Object.assign(params, newFilters)"
             @change="debouncedReload"
             @update:searchQuery="debouncedReload"
@@ -439,7 +440,7 @@ const isColumnVisible = (key: string) => !hiddenColumns.has(key)
 
 const { items: accounts, loading, params, pagination, load, reload, debouncedReload, handlePageChange, handlePageSizeChange } = useTableLoader<Account, any>({
   fetchFn: adminAPI.accounts.list,
-  initialParams: { platform: '', type: '', status: '', search: '' }
+  initialParams: { platform: '', type: '', status: '', group: '', search: '' }
 })
 
 const isAnyModalOpen = computed(() => {

frontend/src/views/admin/RedeemView.vue

@@ -117,9 +117,9 @@
             </span>
           </template>
 
-          <template #cell-used_by="{ value }">
+          <template #cell-used_by="{ value, row }">
             <span class="text-sm text-gray-500 dark:text-dark-400">
-              {{ value ? t('admin.redeem.userPrefix', { id: value }) : '-' }}
+              {{ row.user?.email || (value ? t('admin.redeem.userPrefix', { id: value }) : '-') }}
             </span>
           </template>
 

frontend/src/views/user/UsageView.vue

@@ -233,6 +233,8 @@
                     <span class="font-medium text-amber-600 dark:text-amber-400">{{
                       formatCacheTokens(row.cache_creation_tokens)
                     }}</span>
+                    <span v-if="row.cache_creation_1h_tokens > 0" class="inline-flex items-center rounded px-1 py-px text-[10px] font-medium leading-tight bg-orange-100 text-orange-600 ring-1 ring-inset ring-orange-200 dark:bg-orange-500/20 dark:text-orange-400 dark:ring-orange-500/30">1h</span>
+                    <span v-if="row.cache_ttl_overridden" :title="t('usage.cacheTtlOverriddenHint')" class="inline-flex items-center rounded px-1 py-px text-[10px] font-medium leading-tight bg-rose-100 text-rose-600 ring-1 ring-inset ring-rose-200 dark:bg-rose-500/20 dark:text-rose-400 dark:ring-rose-500/30 cursor-help">R</span>
                   </div>
                 </div>
               </div>
@@ -350,9 +352,36 @@
               <span class="text-gray-400">{{ t('admin.usage.outputTokens') }}</span>
               <span class="font-medium text-white">{{ tokenTooltipData.output_tokens.toLocaleString() }}</span>
             </div>
-            <div v-if="tokenTooltipData && tokenTooltipData.cache_creation_tokens > 0" class="flex items-center justify-between gap-4">
+            <div v-if="tokenTooltipData && tokenTooltipData.cache_creation_tokens > 0">
-              <span class="text-gray-400">{{ t('admin.usage.cacheCreationTokens') }}</span>
+              <!-- 有 5m/1h 明细时，展开显示 -->
-              <span class="font-medium text-white">{{ tokenTooltipData.cache_creation_tokens.toLocaleString() }}</span>
+              <template v-if="tokenTooltipData.cache_creation_5m_tokens > 0 || tokenTooltipData.cache_creation_1h_tokens > 0">
+                <div v-if="tokenTooltipData.cache_creation_5m_tokens > 0" class="flex items-center justify-between gap-4">
+                  <span class="text-gray-400 flex items-center gap-1.5">
+                    {{ t('admin.usage.cacheCreation5mTokens') }}
+                    <span class="inline-flex items-center rounded px-1 py-px text-[10px] font-medium leading-tight bg-amber-500/20 text-amber-400 ring-1 ring-inset ring-amber-500/30">5m</span>
+                  </span>
+                  <span class="font-medium text-white">{{ tokenTooltipData.cache_creation_5m_tokens.toLocaleString() }}</span>
+                </div>
+                <div v-if="tokenTooltipData.cache_creation_1h_tokens > 0" class="flex items-center justify-between gap-4">
+                  <span class="text-gray-400 flex items-center gap-1.5">
+                    {{ t('admin.usage.cacheCreation1hTokens') }}
+                    <span class="inline-flex items-center rounded px-1 py-px text-[10px] font-medium leading-tight bg-orange-500/20 text-orange-400 ring-1 ring-inset ring-orange-500/30">1h</span>
+                  </span>
+                  <span class="font-medium text-white">{{ tokenTooltipData.cache_creation_1h_tokens.toLocaleString() }}</span>
+                </div>
+              </template>
+              <!-- 无明细时，只显示聚合值 -->
+              <div v-else class="flex items-center justify-between gap-4">
+                <span class="text-gray-400">{{ t('admin.usage.cacheCreationTokens') }}</span>
+                <span class="font-medium text-white">{{ tokenTooltipData.cache_creation_tokens.toLocaleString() }}</span>
+              </div>
+            </div>
+            <div v-if="tokenTooltipData && tokenTooltipData.cache_ttl_overridden" class="flex items-center justify-between gap-4">
+              <span class="text-gray-400 flex items-center gap-1.5">
+                {{ t('usage.cacheTtlOverriddenLabel') }}
+                <span class="inline-flex items-center rounded px-1 py-px text-[10px] font-medium leading-tight bg-rose-500/20 text-rose-400 ring-1 ring-inset ring-rose-500/30">R-{{ tokenTooltipData.cache_creation_1h_tokens > 0 ? '5m' : '1H' }}</span>
+              </span>
+              <span class="font-medium text-rose-400">{{ tokenTooltipData.cache_creation_1h_tokens > 0 ? t('usage.cacheTtlOverridden1h') : t('usage.cacheTtlOverridden5m') }}</span>
             </div>
             <div v-if="tokenTooltipData && tokenTooltipData.cache_read_tokens > 0" class="flex items-center justify-between gap-4">
               <span class="text-gray-400">{{ t('admin.usage.cacheReadTokens') }}</span>