Merge pull request #1043 from touwaeriol/pr/antigravity-credits-overages

feat: Antigravity AI Credits overages handling & balance display

backend/internal/pkg/antigravity/client.go

@@ -124,10 +124,68 @@ type IneligibleTier struct {
 type LoadCodeAssistResponse struct {
 	CloudAICompanionProject string            `json:"cloudaicompanionProject"`
 	CurrentTier             *TierInfo         `json:"currentTier,omitempty"`
-	PaidTier                *TierInfo         `json:"paidTier,omitempty"`
+	PaidTier                *PaidTierInfo     `json:"paidTier,omitempty"`
 	IneligibleTiers         []*IneligibleTier `json:"ineligibleTiers,omitempty"`
 }
 
+// PaidTierInfo 付费等级信息，包含 AI Credits 余额。
+type PaidTierInfo struct {
+	ID               string            `json:"id"`
+	Name             string            `json:"name"`
+	Description      string            `json:"description"`
+	AvailableCredits []AvailableCredit `json:"availableCredits,omitempty"`
+}
+
+// UnmarshalJSON 兼容 paidTier 既可能是字符串也可能是对象的情况。
+func (p *PaidTierInfo) UnmarshalJSON(data []byte) error {
+	data = bytes.TrimSpace(data)
+	if len(data) == 0 || string(data) == "null" {
+		return nil
+	}
+	if data[0] == '"' {
+		var id string
+		if err := json.Unmarshal(data, &id); err != nil {
+			return err
+		}
+		p.ID = id
+		return nil
+	}
+	type alias PaidTierInfo
+	var raw alias
+	if err := json.Unmarshal(data, &raw); err != nil {
+		return err
+	}
+	*p = PaidTierInfo(raw)
+	return nil
+}
+
+// AvailableCredit 表示一条 AI Credits 余额记录。
+type AvailableCredit struct {
+	CreditType                  string `json:"creditType,omitempty"`
+	CreditAmount                string `json:"creditAmount,omitempty"`
+	MinimumCreditAmountForUsage string `json:"minimumCreditAmountForUsage,omitempty"`
+}
+
+// GetAmount 将 creditAmount 解析为浮点数。
+func (c *AvailableCredit) GetAmount() float64 {
+	if c.CreditAmount == "" {
+		return 0
+	}
+	var value float64
+	_, _ = fmt.Sscanf(c.CreditAmount, "%f", &value)
+	return value
+}
+
+// GetMinimumAmount 将 minimumCreditAmountForUsage 解析为浮点数。
+func (c *AvailableCredit) GetMinimumAmount() float64 {
+	if c.MinimumCreditAmountForUsage == "" {
+		return 0
+	}
+	var value float64
+	_, _ = fmt.Sscanf(c.MinimumCreditAmountForUsage, "%f", &value)
+	return value
+}
+
 // OnboardUserRequest onboardUser 请求
 type OnboardUserRequest struct {
 	TierID   string `json:"tierId"`
@@ -157,6 +215,14 @@ func (r *LoadCodeAssistResponse) GetTier() string {
 	return ""
 }
 
+// GetAvailableCredits 返回 paid tier 中的 AI Credits 余额列表。
+func (r *LoadCodeAssistResponse) GetAvailableCredits() []AvailableCredit {
+	if r.PaidTier == nil {
+		return nil
+	}
+	return r.PaidTier.AvailableCredits
+}
+
 // Client Antigravity API 客户端
 type Client struct {
 	httpClient *http.Client

backend/internal/pkg/antigravity/client_test.go

@@ -190,7 +190,7 @@ func TestTierInfo_UnmarshalJSON_通过JSON嵌套结构(t *testing.T) {
 func TestGetTier_PaidTier优先(t *testing.T) {
 	resp := &LoadCodeAssistResponse{
 		CurrentTier: &TierInfo{ID: "free-tier"},
-		PaidTier:    &TierInfo{ID: "g1-pro-tier"},
+		PaidTier:    &PaidTierInfo{ID: "g1-pro-tier"},
 	}
 	if got := resp.GetTier(); got != "g1-pro-tier" {
 		t.Errorf("应返回 paidTier: got %s", got)
@@ -209,7 +209,7 @@ func TestGetTier_回退到CurrentTier(t *testing.T) {
 func TestGetTier_PaidTier为空ID(t *testing.T) {
 	resp := &LoadCodeAssistResponse{
 		CurrentTier: &TierInfo{ID: "free-tier"},
-		PaidTier:    &TierInfo{ID: ""},
+		PaidTier:    &PaidTierInfo{ID: ""},
 	}
 	// paidTier.ID 为空时应回退到 currentTier
 	if got := resp.GetTier(); got != "free-tier" {
@@ -217,6 +217,32 @@ func TestGetTier_PaidTier为空ID(t *testing.T) {
 	}
 }
 
+func TestGetAvailableCredits(t *testing.T) {
+	resp := &LoadCodeAssistResponse{
+		PaidTier: &PaidTierInfo{
+			ID: "g1-pro-tier",
+			AvailableCredits: []AvailableCredit{
+				{
+					CreditType:                  "GOOGLE_ONE_AI",
+					CreditAmount:                "25",
+					MinimumCreditAmountForUsage: "5",
+				},
+			},
+		},
+	}
+
+	credits := resp.GetAvailableCredits()
+	if len(credits) != 1 {
+		t.Fatalf("AI Credits 数量不匹配: got %d", len(credits))
+	}
+	if credits[0].GetAmount() != 25 {
+		t.Errorf("CreditAmount 解析不正确: got %v", credits[0].GetAmount())
+	}
+	if credits[0].GetMinimumAmount() != 5 {
+		t.Errorf("MinimumCreditAmountForUsage 解析不正确: got %v", credits[0].GetMinimumAmount())
+	}
+}
+
 func TestGetTier_两者都为nil(t *testing.T) {
 	resp := &LoadCodeAssistResponse{}
 	if got := resp.GetTier(); got != "" {

backend/internal/service/account.go

@@ -901,6 +901,22 @@ func (a *Account) IsMixedSchedulingEnabled() bool {
 	return false
 }
 
+// IsOveragesEnabled 检查 Antigravity 账号是否启用 AI Credits 超量请求。
+func (a *Account) IsOveragesEnabled() bool {
+	if a.Platform != PlatformAntigravity {
+		return false
+	}
+	if a.Extra == nil {
+		return false
+	}
+	if v, ok := a.Extra["allow_overages"]; ok {
+		if enabled, ok := v.(bool); ok {
+			return enabled
+		}
+	}
+	return false
+}
+
 // IsOpenAIPassthroughEnabled 返回 OpenAI 账号是否启用“自动透传（仅替换认证）”。
 //
 // 新字段：accounts.extra.openai_passthrough。

backend/internal/service/account_usage_service.go

@@ -166,6 +166,13 @@ type AntigravityModelDetail struct {
 	SupportedMimeTypes map[string]bool `json:"supported_mime_types,omitempty"`
 }
 
+// AICredit 表示 Antigravity 账号的 AI Credits 余额信息。
+type AICredit struct {
+	CreditType     string  `json:"credit_type,omitempty"`
+	Amount         float64 `json:"amount,omitempty"`
+	MinimumBalance float64 `json:"minimum_balance,omitempty"`
+}
+
 // UsageInfo 账号使用量信息
 type UsageInfo struct {
 	UpdatedAt          *time.Time     `json:"updated_at,omitempty"`           // 更新时间
@@ -189,6 +196,9 @@ type UsageInfo struct {
 	// Antigravity 模型详细能力信息（与 antigravity_quota 同 key）
 	AntigravityQuotaDetails map[string]*AntigravityModelDetail `json:"antigravity_quota_details,omitempty"`
 
+	// Antigravity AI Credits 余额
+	AICredits []AICredit `json:"ai_credits,omitempty"`
+
 	// Antigravity 废弃模型转发规则 (old_model_id -> new_model_id)
 	ModelForwardingRules map[string]string `json:"model_forwarding_rules,omitempty"`
 

backend/internal/service/admin_service.go

@@ -368,6 +368,10 @@ type ProxyExitInfoProber interface {
 	ProbeProxy(ctx context.Context, proxyURL string) (*ProxyExitInfo, int64, error)
 }
 
+type groupExistenceBatchReader interface {
+	ExistsByIDs(ctx context.Context, ids []int64) (map[int64]bool, error)
+}
+
 type proxyQualityTarget struct {
 	Target          string
 	URL             string
@@ -445,10 +449,6 @@ type userGroupRateBatchReader interface {
 	GetByUserIDs(ctx context.Context, userIDs []int64) (map[int64]map[int64]float64, error)
 }
 
-type groupExistenceBatchReader interface {
-	ExistsByIDs(ctx context.Context, ids []int64) (map[int64]bool, error)
-}
-
 // NewAdminService creates a new AdminService
 func NewAdminService(
 	userRepo UserRepository,
@@ -1516,6 +1516,7 @@ func (s *adminServiceImpl) UpdateAccount(ctx context.Context, id int64, input *U
 	if err != nil {
 		return nil, err
 	}
+	wasOveragesEnabled := account.IsOveragesEnabled()
 
 	if input.Name != "" {
 		account.Name = input.Name
@@ -1537,6 +1538,17 @@ func (s *adminServiceImpl) UpdateAccount(ctx context.Context, id int64, input *U
 			}
 		}
 		account.Extra = input.Extra
+		if account.Platform == PlatformAntigravity && wasOveragesEnabled && !account.IsOveragesEnabled() {
+			delete(account.Extra, "antigravity_credits_overages") // 清理旧版 overages 运行态
+			// 清除 AICredits 限流 key
+			if rawLimits, ok := account.Extra[modelRateLimitsKey].(map[string]any); ok {
+				delete(rawLimits, creditsExhaustedKey)
+			}
+		}
+		if account.Platform == PlatformAntigravity && !wasOveragesEnabled && account.IsOveragesEnabled() {
+			delete(account.Extra, modelRateLimitsKey)
+			delete(account.Extra, "antigravity_credits_overages") // 清理旧版 overages 运行态
+		}
 		// 校验并预计算固定时间重置的下次重置时间
 		if err := ValidateQuotaResetConfig(account.Extra); err != nil {
 			return nil, err

backend/internal/service/admin_service_overages_test.go

@@ -0,0 +1,123 @@
+//go:build unit
+
+package service
+
+import (
+	"context"
+	"testing"
+	"time"
+
+	"github.com/stretchr/testify/require"
+)
+
+type updateAccountOveragesRepoStub struct {
+	mockAccountRepoForGemini
+	account     *Account
+	updateCalls int
+}
+
+func (r *updateAccountOveragesRepoStub) GetByID(ctx context.Context, id int64) (*Account, error) {
+	return r.account, nil
+}
+
+func (r *updateAccountOveragesRepoStub) Update(ctx context.Context, account *Account) error {
+	r.updateCalls++
+	r.account = account
+	return nil
+}
+
+func TestUpdateAccount_DisableOveragesClearsAICreditsKey(t *testing.T) {
+	accountID := int64(101)
+	repo := &updateAccountOveragesRepoStub{
+		account: &Account{
+			ID:       accountID,
+			Platform: PlatformAntigravity,
+			Type:     AccountTypeOAuth,
+			Status:   StatusActive,
+			Extra: map[string]any{
+				"allow_overages":   true,
+				"mixed_scheduling": true,
+				modelRateLimitsKey: map[string]any{
+					"claude-sonnet-4-5": map[string]any{
+						"rate_limited_at":     "2026-03-15T00:00:00Z",
+						"rate_limit_reset_at": "2099-03-15T00:00:00Z",
+					},
+					creditsExhaustedKey: map[string]any{
+						"rate_limited_at":     "2026-03-15T00:00:00Z",
+						"rate_limit_reset_at": time.Now().Add(5 * time.Hour).UTC().Format(time.RFC3339),
+					},
+				},
+			},
+		},
+	}
+
+	svc := &adminServiceImpl{accountRepo: repo}
+	updated, err := svc.UpdateAccount(context.Background(), accountID, &UpdateAccountInput{
+		Extra: map[string]any{
+			"mixed_scheduling": true,
+			modelRateLimitsKey: map[string]any{
+				"claude-sonnet-4-5": map[string]any{
+					"rate_limited_at":     "2026-03-15T00:00:00Z",
+					"rate_limit_reset_at": "2099-03-15T00:00:00Z",
+				},
+				creditsExhaustedKey: map[string]any{
+					"rate_limited_at":     "2026-03-15T00:00:00Z",
+					"rate_limit_reset_at": time.Now().Add(5 * time.Hour).UTC().Format(time.RFC3339),
+				},
+			},
+		},
+	})
+
+	require.NoError(t, err)
+	require.NotNil(t, updated)
+	require.Equal(t, 1, repo.updateCalls)
+	require.False(t, updated.IsOveragesEnabled())
+
+	// 关闭 overages 后，AICredits key 应被清除
+	rawLimits, ok := repo.account.Extra[modelRateLimitsKey].(map[string]any)
+	if ok {
+		_, exists := rawLimits[creditsExhaustedKey]
+		require.False(t, exists, "关闭 overages 时应清除 AICredits 限流 key")
+	}
+	// 普通模型限流应保留
+	require.True(t, ok)
+	_, exists := rawLimits["claude-sonnet-4-5"]
+	require.True(t, exists, "普通模型限流应保留")
+}
+
+func TestUpdateAccount_EnableOveragesClearsModelRateLimitsBeforePersist(t *testing.T) {
+	accountID := int64(102)
+	repo := &updateAccountOveragesRepoStub{
+		account: &Account{
+			ID:       accountID,
+			Platform: PlatformAntigravity,
+			Type:     AccountTypeOAuth,
+			Status:   StatusActive,
+			Extra: map[string]any{
+				"mixed_scheduling": true,
+				modelRateLimitsKey: map[string]any{
+					"claude-sonnet-4-5": map[string]any{
+						"rate_limited_at":     "2026-03-15T00:00:00Z",
+						"rate_limit_reset_at": "2099-03-15T00:00:00Z",
+					},
+				},
+			},
+		},
+	}
+
+	svc := &adminServiceImpl{accountRepo: repo}
+	updated, err := svc.UpdateAccount(context.Background(), accountID, &UpdateAccountInput{
+		Extra: map[string]any{
+			"mixed_scheduling": true,
+			"allow_overages":   true,
+		},
+	})
+
+	require.NoError(t, err)
+	require.NotNil(t, updated)
+	require.Equal(t, 1, repo.updateCalls)
+	require.True(t, updated.IsOveragesEnabled())
+
+	_, exists := repo.account.Extra[modelRateLimitsKey]
+	require.False(t, exists, "开启 overages 时应在持久化前清掉旧模型限流")
+}

backend/internal/service/antigravity_credits_overages.go

@@ -0,0 +1,234 @@
+package service
+
+import (
+	"context"
+	"encoding/json"
+	"io"
+	"net/http"
+	"strings"
+	"time"
+
+	"github.com/Wei-Shaw/sub2api/internal/pkg/antigravity"
+	"github.com/Wei-Shaw/sub2api/internal/pkg/logger"
+)
+
+const (
+	// creditsExhaustedKey 是 model_rate_limits 中标记积分耗尽的特殊 key。
+	// 与普通模型限流完全同构：通过 SetModelRateLimit / isRateLimitActiveForKey 读写。
+	creditsExhaustedKey      = "AICredits"
+	creditsExhaustedDuration = 5 * time.Hour
+)
+
+type antigravity429Category string
+
+const (
+	antigravity429Unknown        antigravity429Category = "unknown"
+	antigravity429RateLimited    antigravity429Category = "rate_limited"
+	antigravity429QuotaExhausted antigravity429Category = "quota_exhausted"
+)
+
+var (
+	antigravityQuotaExhaustedKeywords = []string{
+		"quota_exhausted",
+		"quota exhausted",
+	}
+
+	creditsExhaustedKeywords = []string{
+		"google_one_ai",
+		"insufficient credit",
+		"insufficient credits",
+		"not enough credit",
+		"not enough credits",
+		"credit exhausted",
+		"credits exhausted",
+		"credit balance",
+		"minimumcreditamountforusage",
+		"minimum credit amount for usage",
+		"minimum credit",
+	}
+)
+
+// isCreditsExhausted 检查账号的 AICredits 限流 key 是否生效（积分是否耗尽）。
+func (a *Account) isCreditsExhausted() bool {
+	if a == nil {
+		return false
+	}
+	return a.isRateLimitActiveForKey(creditsExhaustedKey)
+}
+
+// setCreditsExhausted 标记账号积分耗尽：写入 model_rate_limits["AICredits"] + 更新缓存。
+func (s *AntigravityGatewayService) setCreditsExhausted(ctx context.Context, account *Account) {
+	if account == nil || account.ID == 0 {
+		return
+	}
+	resetAt := time.Now().Add(creditsExhaustedDuration)
+	if err := s.accountRepo.SetModelRateLimit(ctx, account.ID, creditsExhaustedKey, resetAt); err != nil {
+		logger.LegacyPrintf("service.antigravity_gateway", "set credits exhausted failed: account=%d err=%v", account.ID, err)
+		return
+	}
+	s.updateAccountModelRateLimitInCache(ctx, account, creditsExhaustedKey, resetAt)
+	logger.LegacyPrintf("service.antigravity_gateway", "credits_exhausted_marked account=%d reset_at=%s",
+		account.ID, resetAt.UTC().Format(time.RFC3339))
+}
+
+// clearCreditsExhausted 清除账号的 AICredits 限流 key。
+func (s *AntigravityGatewayService) clearCreditsExhausted(ctx context.Context, account *Account) {
+	if account == nil || account.ID == 0 || account.Extra == nil {
+		return
+	}
+	rawLimits, ok := account.Extra[modelRateLimitsKey].(map[string]any)
+	if !ok {
+		return
+	}
+	if _, exists := rawLimits[creditsExhaustedKey]; !exists {
+		return
+	}
+	delete(rawLimits, creditsExhaustedKey)
+	account.Extra[modelRateLimitsKey] = rawLimits
+	if err := s.accountRepo.UpdateExtra(ctx, account.ID, map[string]any{
+		modelRateLimitsKey: rawLimits,
+	}); err != nil {
+		logger.LegacyPrintf("service.antigravity_gateway", "clear credits exhausted failed: account=%d err=%v", account.ID, err)
+	}
+}
+
+// classifyAntigravity429 将 Antigravity 的 429 响应归类为配额耗尽、限流或未知。
+func classifyAntigravity429(body []byte) antigravity429Category {
+	if len(body) == 0 {
+		return antigravity429Unknown
+	}
+	lowerBody := strings.ToLower(string(body))
+	for _, keyword := range antigravityQuotaExhaustedKeywords {
+		if strings.Contains(lowerBody, keyword) {
+			return antigravity429QuotaExhausted
+		}
+	}
+	if info := parseAntigravitySmartRetryInfo(body); info != nil && !info.IsModelCapacityExhausted {
+		return antigravity429RateLimited
+	}
+	return antigravity429Unknown
+}
+
+// injectEnabledCreditTypes 在已序列化的 v1internal JSON body 中注入 AI Credits 类型。
+func injectEnabledCreditTypes(body []byte) []byte {
+	var payload map[string]any
+	if err := json.Unmarshal(body, &payload); err != nil {
+		return nil
+	}
+	payload["enabledCreditTypes"] = []string{"GOOGLE_ONE_AI"}
+	result, err := json.Marshal(payload)
+	if err != nil {
+		return nil
+	}
+	return result
+}
+
+// resolveCreditsOveragesModelKey 解析当前请求对应的 overages 状态模型 key。
+func resolveCreditsOveragesModelKey(ctx context.Context, account *Account, upstreamModelName, requestedModel string) string {
+	modelKey := strings.TrimSpace(upstreamModelName)
+	if modelKey != "" {
+		return modelKey
+	}
+	if account == nil {
+		return ""
+	}
+	modelKey = resolveFinalAntigravityModelKey(ctx, account, requestedModel)
+	if strings.TrimSpace(modelKey) != "" {
+		return modelKey
+	}
+	return resolveAntigravityModelKey(requestedModel)
+}
+
+// shouldMarkCreditsExhausted 判断一次 credits 请求失败是否应标记为 credits 耗尽。
+func shouldMarkCreditsExhausted(resp *http.Response, respBody []byte, reqErr error) bool {
+	if reqErr != nil || resp == nil {
+		return false
+	}
+	if resp.StatusCode >= 500 || resp.StatusCode == http.StatusRequestTimeout {
+		return false
+	}
+	if isURLLevelRateLimit(respBody) {
+		return false
+	}
+	if info := parseAntigravitySmartRetryInfo(respBody); info != nil {
+		return false
+	}
+	bodyLower := strings.ToLower(string(respBody))
+	for _, keyword := range creditsExhaustedKeywords {
+		if strings.Contains(bodyLower, keyword) {
+			return true
+		}
+	}
+	return false
+}
+
+type creditsOveragesRetryResult struct {
+	handled bool
+	resp    *http.Response
+}
+
+// attemptCreditsOveragesRetry 在确认免费配额耗尽后，尝试注入 AI Credits 继续请求。
+func (s *AntigravityGatewayService) attemptCreditsOveragesRetry(
+	p antigravityRetryLoopParams,
+	baseURL string,
+	modelName string,
+	waitDuration time.Duration,
+	originalStatusCode int,
+	respBody []byte,
+) *creditsOveragesRetryResult {
+	creditsBody := injectEnabledCreditTypes(p.body)
+	if creditsBody == nil {
+		return &creditsOveragesRetryResult{handled: false}
+	}
+	modelKey := resolveCreditsOveragesModelKey(p.ctx, p.account, modelName, p.requestedModel)
+	logger.LegacyPrintf("service.antigravity_gateway", "%s status=429 credit_overages_retry model=%s account=%d (injecting enabledCreditTypes)",
+		p.prefix, modelKey, p.account.ID)
+
+	creditsReq, err := antigravity.NewAPIRequestWithURL(p.ctx, baseURL, p.action, p.accessToken, creditsBody)
+	if err != nil {
+		logger.LegacyPrintf("service.antigravity_gateway", "%s credit_overages_failed model=%s account=%d build_request_err=%v",
+			p.prefix, modelKey, p.account.ID, err)
+		return &creditsOveragesRetryResult{handled: true}
+	}
+
+	creditsResp, err := p.httpUpstream.Do(creditsReq, p.proxyURL, p.account.ID, p.account.Concurrency)
+	if err == nil && creditsResp != nil && creditsResp.StatusCode < 400 {
+		s.clearCreditsExhausted(p.ctx, p.account)
+		logger.LegacyPrintf("service.antigravity_gateway", "%s status=%d credit_overages_success model=%s account=%d",
+			p.prefix, creditsResp.StatusCode, modelKey, p.account.ID)
+		return &creditsOveragesRetryResult{handled: true, resp: creditsResp}
+	}
+
+	s.handleCreditsRetryFailure(p.ctx, p.prefix, modelKey, p.account, creditsResp, err)
+	return &creditsOveragesRetryResult{handled: true}
+}
+
+func (s *AntigravityGatewayService) handleCreditsRetryFailure(
+	ctx context.Context,
+	prefix string,
+	modelKey string,
+	account *Account,
+	creditsResp *http.Response,
+	reqErr error,
+) {
+	var creditsRespBody []byte
+	creditsStatusCode := 0
+	if creditsResp != nil {
+		creditsStatusCode = creditsResp.StatusCode
+		if creditsResp.Body != nil {
+			creditsRespBody, _ = io.ReadAll(io.LimitReader(creditsResp.Body, 64<<10))
+			_ = creditsResp.Body.Close()
+		}
+	}
+
+	if shouldMarkCreditsExhausted(creditsResp, creditsRespBody, reqErr) && account != nil {
+		s.setCreditsExhausted(ctx, account)
+		logger.LegacyPrintf("service.antigravity_gateway", "%s credit_overages_failed model=%s account=%d marked_exhausted=true status=%d body=%s",
+			prefix, modelKey, account.ID, creditsStatusCode, truncateForLog(creditsRespBody, 200))
+		return
+	}
+	if account != nil {
+		logger.LegacyPrintf("service.antigravity_gateway", "%s credit_overages_failed model=%s account=%d marked_exhausted=false status=%d err=%v body=%s",
+			prefix, modelKey, account.ID, creditsStatusCode, reqErr, truncateForLog(creditsRespBody, 200))
+	}
+}

backend/internal/service/antigravity_credits_overages_test.go

@@ -0,0 +1,538 @@
+//go:build unit
+
+package service
+
+import (
+	"bytes"
+	"context"
+	"io"
+	"net/http"
+	"strings"
+	"testing"
+	"time"
+
+	"github.com/Wei-Shaw/sub2api/internal/pkg/antigravity"
+	"github.com/stretchr/testify/require"
+)
+
+func TestClassifyAntigravity429(t *testing.T) {
+	t.Run("明确配额耗尽", func(t *testing.T) {
+		body := []byte(`{"error":{"status":"RESOURCE_EXHAUSTED","message":"QUOTA_EXHAUSTED"}}`)
+		require.Equal(t, antigravity429QuotaExhausted, classifyAntigravity429(body))
+	})
+
+	t.Run("结构化限流", func(t *testing.T) {
+		body := []byte(`{
+			"error": {
+				"status": "RESOURCE_EXHAUSTED",
+				"details": [
+					{"@type": "type.googleapis.com/google.rpc.ErrorInfo", "metadata": {"model": "claude-sonnet-4-5"}, "reason": "RATE_LIMIT_EXCEEDED"},
+					{"@type": "type.googleapis.com/google.rpc.RetryInfo", "retryDelay": "0.5s"}
+				]
+			}
+		}`)
+		require.Equal(t, antigravity429RateLimited, classifyAntigravity429(body))
+	})
+
+	t.Run("未知429", func(t *testing.T) {
+		body := []byte(`{"error":{"message":"too many requests"}}`)
+		require.Equal(t, antigravity429Unknown, classifyAntigravity429(body))
+	})
+}
+
+func TestIsCreditsExhausted_UsesAICreditsKey(t *testing.T) {
+	t.Run("无 AICredits key 则积分可用", func(t *testing.T) {
+		account := &Account{
+			ID:       1,
+			Platform: PlatformAntigravity,
+			Extra: map[string]any{
+				"allow_overages": true,
+			},
+		}
+		require.False(t, account.isCreditsExhausted())
+	})
+
+	t.Run("AICredits key 生效则积分耗尽", func(t *testing.T) {
+		account := &Account{
+			ID:       2,
+			Platform: PlatformAntigravity,
+			Extra: map[string]any{
+				"allow_overages": true,
+				modelRateLimitsKey: map[string]any{
+					creditsExhaustedKey: map[string]any{
+						"rate_limited_at":     time.Now().UTC().Format(time.RFC3339),
+						"rate_limit_reset_at": time.Now().Add(5 * time.Hour).UTC().Format(time.RFC3339),
+					},
+				},
+			},
+		}
+		require.True(t, account.isCreditsExhausted())
+	})
+
+	t.Run("AICredits key 过期则积分可用", func(t *testing.T) {
+		account := &Account{
+			ID:       3,
+			Platform: PlatformAntigravity,
+			Extra: map[string]any{
+				"allow_overages": true,
+				modelRateLimitsKey: map[string]any{
+					creditsExhaustedKey: map[string]any{
+						"rate_limited_at":     time.Now().Add(-6 * time.Hour).UTC().Format(time.RFC3339),
+						"rate_limit_reset_at": time.Now().Add(-1 * time.Hour).UTC().Format(time.RFC3339),
+					},
+				},
+			},
+		}
+		require.False(t, account.isCreditsExhausted())
+	})
+}
+
+func TestHandleSmartRetry_QuotaExhausted_UsesCreditsAndStoresIndependentState(t *testing.T) {
+	successResp := &http.Response{
+		StatusCode: http.StatusOK,
+		Header:     http.Header{},
+		Body:       io.NopCloser(strings.NewReader(`{"ok":true}`)),
+	}
+	upstream := &mockSmartRetryUpstream{
+		responses: []*http.Response{successResp},
+		errors:    []error{nil},
+	}
+	repo := &stubAntigravityAccountRepo{}
+	account := &Account{
+		ID:       101,
+		Name:     "acc-101",
+		Type:     AccountTypeOAuth,
+		Platform: PlatformAntigravity,
+		Extra: map[string]any{
+			"allow_overages": true,
+		},
+		Credentials: map[string]any{
+			"model_mapping": map[string]any{
+				"claude-opus-4-6": "claude-sonnet-4-5",
+			},
+		},
+	}
+
+	respBody := []byte(`{"error":{"status":"RESOURCE_EXHAUSTED","message":"QUOTA_EXHAUSTED"}}`)
+	resp := &http.Response{
+		StatusCode: http.StatusTooManyRequests,
+		Header:     http.Header{},
+		Body:       io.NopCloser(bytes.NewReader(respBody)),
+	}
+	params := antigravityRetryLoopParams{
+		ctx:            context.Background(),
+		prefix:         "[test]",
+		account:        account,
+		accessToken:    "token",
+		action:         "generateContent",
+		body:           []byte(`{"model":"claude-opus-4-6","request":{}}`),
+		httpUpstream:   upstream,
+		accountRepo:    repo,
+		requestedModel: "claude-opus-4-6",
+		handleError: func(ctx context.Context, prefix string, account *Account, statusCode int, headers http.Header, body []byte, requestedModel string, groupID int64, sessionHash string, isStickySession bool) *handleModelRateLimitResult {
+			return nil
+		},
+	}
+
+	svc := &AntigravityGatewayService{}
+	result := svc.handleSmartRetry(params, resp, respBody, "https://ag-1.test", 0, []string{"https://ag-1.test"})
+
+	require.NotNil(t, result)
+	require.Equal(t, smartRetryActionBreakWithResp, result.action)
+	require.NotNil(t, result.resp)
+	require.Nil(t, result.switchError)
+	require.Len(t, upstream.requestBodies, 1)
+	require.Contains(t, string(upstream.requestBodies[0]), "enabledCreditTypes")
+	require.Empty(t, repo.modelRateLimitCalls, "overages 成功后不应写入普通 model_rate_limits")
+}
+
+func TestHandleSmartRetry_RateLimited_DoesNotUseCredits(t *testing.T) {
+	successResp := &http.Response{
+		StatusCode: http.StatusOK,
+		Header:     http.Header{},
+		Body:       io.NopCloser(strings.NewReader(`{"ok":true}`)),
+	}
+	upstream := &mockSmartRetryUpstream{
+		responses: []*http.Response{successResp},
+		errors:    []error{nil},
+	}
+	repo := &stubAntigravityAccountRepo{}
+	account := &Account{
+		ID:       102,
+		Name:     "acc-102",
+		Type:     AccountTypeOAuth,
+		Platform: PlatformAntigravity,
+		Extra: map[string]any{
+			"allow_overages": true,
+		},
+	}
+
+	respBody := []byte(`{
+		"error": {
+			"status": "RESOURCE_EXHAUSTED",
+			"details": [
+				{"@type": "type.googleapis.com/google.rpc.ErrorInfo", "metadata": {"model": "claude-sonnet-4-5"}, "reason": "RATE_LIMIT_EXCEEDED"},
+				{"@type": "type.googleapis.com/google.rpc.RetryInfo", "retryDelay": "0.1s"}
+			]
+		}
+	}`)
+	resp := &http.Response{
+		StatusCode: http.StatusTooManyRequests,
+		Header:     http.Header{},
+		Body:       io.NopCloser(bytes.NewReader(respBody)),
+	}
+	params := antigravityRetryLoopParams{
+		ctx:          context.Background(),
+		prefix:       "[test]",
+		account:      account,
+		accessToken:  "token",
+		action:       "generateContent",
+		body:         []byte(`{"model":"claude-sonnet-4-5","request":{}}`),
+		httpUpstream: upstream,
+		accountRepo:  repo,
+		handleError: func(ctx context.Context, prefix string, account *Account, statusCode int, headers http.Header, body []byte, requestedModel string, groupID int64, sessionHash string, isStickySession bool) *handleModelRateLimitResult {
+			return nil
+		},
+	}
+
+	svc := &AntigravityGatewayService{}
+	result := svc.handleSmartRetry(params, resp, respBody, "https://ag-1.test", 0, []string{"https://ag-1.test"})
+
+	require.NotNil(t, result)
+	require.Equal(t, smartRetryActionBreakWithResp, result.action)
+	require.NotNil(t, result.resp)
+	require.Len(t, upstream.requestBodies, 1)
+	require.NotContains(t, string(upstream.requestBodies[0]), "enabledCreditTypes")
+	require.Empty(t, repo.extraUpdateCalls)
+	require.Empty(t, repo.modelRateLimitCalls)
+}
+
+func TestAntigravityRetryLoop_ModelRateLimited_InjectsCredits(t *testing.T) {
+	oldBaseURLs := append([]string(nil), antigravity.BaseURLs...)
+	oldAvailability := antigravity.DefaultURLAvailability
+	defer func() {
+		antigravity.BaseURLs = oldBaseURLs
+		antigravity.DefaultURLAvailability = oldAvailability
+	}()
+
+	antigravity.BaseURLs = []string{"https://ag-1.test"}
+	antigravity.DefaultURLAvailability = antigravity.NewURLAvailability(time.Minute)
+
+	upstream := &queuedHTTPUpstreamStub{
+		responses: []*http.Response{
+			{
+				StatusCode: http.StatusOK,
+				Header:     http.Header{},
+				Body:       io.NopCloser(strings.NewReader(`{"ok":true}`)),
+			},
+		},
+		errors: []error{nil},
+	}
+	// 模型已限流 + overages 启用 + 无 AICredits key → 应直接注入积分
+	account := &Account{
+		ID:          103,
+		Name:        "acc-103",
+		Type:        AccountTypeOAuth,
+		Platform:    PlatformAntigravity,
+		Status:      StatusActive,
+		Schedulable: true,
+		Extra: map[string]any{
+			"allow_overages": true,
+			modelRateLimitsKey: map[string]any{
+				"claude-sonnet-4-5": map[string]any{
+					"rate_limited_at":     time.Now().UTC().Format(time.RFC3339),
+					"rate_limit_reset_at": time.Now().Add(30 * time.Minute).UTC().Format(time.RFC3339),
+				},
+			},
+		},
+	}
+
+	svc := &AntigravityGatewayService{}
+	result, err := svc.antigravityRetryLoop(antigravityRetryLoopParams{
+		ctx:            context.Background(),
+		prefix:         "[test]",
+		account:        account,
+		accessToken:    "token",
+		action:         "generateContent",
+		body:           []byte(`{"model":"claude-sonnet-4-5","request":{}}`),
+		httpUpstream:   upstream,
+		requestedModel: "claude-sonnet-4-5",
+		handleError: func(ctx context.Context, prefix string, account *Account, statusCode int, headers http.Header, body []byte, requestedModel string, groupID int64, sessionHash string, isStickySession bool) *handleModelRateLimitResult {
+			return nil
+		},
+	})
+
+	require.NoError(t, err)
+	require.NotNil(t, result)
+	require.Len(t, upstream.requestBodies, 1)
+	require.Contains(t, string(upstream.requestBodies[0]), "enabledCreditTypes")
+}
+
+func TestAntigravityRetryLoop_CreditsExhausted_DoesNotInject(t *testing.T) {
+	oldBaseURLs := append([]string(nil), antigravity.BaseURLs...)
+	oldAvailability := antigravity.DefaultURLAvailability
+	defer func() {
+		antigravity.BaseURLs = oldBaseURLs
+		antigravity.DefaultURLAvailability = oldAvailability
+	}()
+
+	antigravity.BaseURLs = []string{"https://ag-1.test"}
+	antigravity.DefaultURLAvailability = antigravity.NewURLAvailability(time.Minute)
+
+	// 模型限流 + overages 启用 + AICredits key 生效 → 不应注入积分，应切号
+	account := &Account{
+		ID:          104,
+		Name:        "acc-104",
+		Type:        AccountTypeOAuth,
+		Platform:    PlatformAntigravity,
+		Status:      StatusActive,
+		Schedulable: true,
+		Extra: map[string]any{
+			"allow_overages": true,
+			modelRateLimitsKey: map[string]any{
+				"claude-sonnet-4-5": map[string]any{
+					"rate_limited_at":     time.Now().UTC().Format(time.RFC3339),
+					"rate_limit_reset_at": time.Now().Add(30 * time.Minute).UTC().Format(time.RFC3339),
+				},
+				creditsExhaustedKey: map[string]any{
+					"rate_limited_at":     time.Now().UTC().Format(time.RFC3339),
+					"rate_limit_reset_at": time.Now().Add(5 * time.Hour).UTC().Format(time.RFC3339),
+				},
+			},
+		},
+	}
+
+	svc := &AntigravityGatewayService{}
+	_, err := svc.antigravityRetryLoop(antigravityRetryLoopParams{
+		ctx:            context.Background(),
+		prefix:         "[test]",
+		account:        account,
+		accessToken:    "token",
+		action:         "generateContent",
+		body:           []byte(`{"model":"claude-sonnet-4-5","request":{}}`),
+		requestedModel: "claude-sonnet-4-5",
+		handleError: func(ctx context.Context, prefix string, account *Account, statusCode int, headers http.Header, body []byte, requestedModel string, groupID int64, sessionHash string, isStickySession bool) *handleModelRateLimitResult {
+			return nil
+		},
+	})
+
+	// 模型限流 + 积分耗尽 → 应触发切号错误
+	require.Error(t, err)
+	var switchErr *AntigravityAccountSwitchError
+	require.ErrorAs(t, err, &switchErr)
+}
+
+func TestAntigravityRetryLoop_CreditErrorMarksExhausted(t *testing.T) {
+	oldBaseURLs := append([]string(nil), antigravity.BaseURLs...)
+	oldAvailability := antigravity.DefaultURLAvailability
+	defer func() {
+		antigravity.BaseURLs = oldBaseURLs
+		antigravity.DefaultURLAvailability = oldAvailability
+	}()
+
+	antigravity.BaseURLs = []string{"https://ag-1.test"}
+	antigravity.DefaultURLAvailability = antigravity.NewURLAvailability(time.Minute)
+
+	repo := &stubAntigravityAccountRepo{}
+	upstream := &queuedHTTPUpstreamStub{
+		responses: []*http.Response{
+			{
+				StatusCode: http.StatusForbidden,
+				Header:     http.Header{},
+				Body:       io.NopCloser(strings.NewReader(`{"error":{"message":"Insufficient GOOGLE_ONE_AI credits"}}`)),
+			},
+		},
+		errors: []error{nil},
+	}
+	// 模型限流 + overages 启用 + 积分可用 → 注入积分但上游返回积分不足
+	account := &Account{
+		ID:          105,
+		Name:        "acc-105",
+		Type:        AccountTypeOAuth,
+		Platform:    PlatformAntigravity,
+		Status:      StatusActive,
+		Schedulable: true,
+		Extra: map[string]any{
+			"allow_overages": true,
+			modelRateLimitsKey: map[string]any{
+				"claude-sonnet-4-5": map[string]any{
+					"rate_limited_at":     time.Now().UTC().Format(time.RFC3339),
+					"rate_limit_reset_at": time.Now().Add(30 * time.Minute).UTC().Format(time.RFC3339),
+				},
+			},
+		},
+	}
+
+	svc := &AntigravityGatewayService{accountRepo: repo}
+	result, err := svc.antigravityRetryLoop(antigravityRetryLoopParams{
+		ctx:            context.Background(),
+		prefix:         "[test]",
+		account:        account,
+		accessToken:    "token",
+		action:         "generateContent",
+		body:           []byte(`{"model":"claude-sonnet-4-5","request":{}}`),
+		httpUpstream:   upstream,
+		accountRepo:    repo,
+		requestedModel: "claude-sonnet-4-5",
+		handleError: func(ctx context.Context, prefix string, account *Account, statusCode int, headers http.Header, body []byte, requestedModel string, groupID int64, sessionHash string, isStickySession bool) *handleModelRateLimitResult {
+			return nil
+		},
+	})
+
+	require.NoError(t, err)
+	require.NotNil(t, result)
+	// 验证 AICredits key 已通过 SetModelRateLimit 写入数据库
+	require.Len(t, repo.modelRateLimitCalls, 1, "应通过 SetModelRateLimit 写入 AICredits key")
+	require.Equal(t, creditsExhaustedKey, repo.modelRateLimitCalls[0].modelKey)
+}
+
+func TestShouldMarkCreditsExhausted(t *testing.T) {
+	t.Run("reqErr 不为 nil 时不标记", func(t *testing.T) {
+		resp := &http.Response{StatusCode: http.StatusForbidden}
+		require.False(t, shouldMarkCreditsExhausted(resp, []byte(`{"error":"Insufficient credits"}`), io.ErrUnexpectedEOF))
+	})
+
+	t.Run("resp 为 nil 时不标记", func(t *testing.T) {
+		require.False(t, shouldMarkCreditsExhausted(nil, []byte(`{"error":"Insufficient credits"}`), nil))
+	})
+
+	t.Run("5xx 响应不标记", func(t *testing.T) {
+		resp := &http.Response{StatusCode: http.StatusInternalServerError}
+		require.False(t, shouldMarkCreditsExhausted(resp, []byte(`{"error":"Insufficient credits"}`), nil))
+	})
+
+	t.Run("408 RequestTimeout 不标记", func(t *testing.T) {
+		resp := &http.Response{StatusCode: http.StatusRequestTimeout}
+		require.False(t, shouldMarkCreditsExhausted(resp, []byte(`{"error":"Insufficient credits"}`), nil))
+	})
+
+	t.Run("URL 级限流不标记", func(t *testing.T) {
+		resp := &http.Response{StatusCode: http.StatusTooManyRequests}
+		body := []byte(`{"error":{"message":"Resource has been exhausted"}}`)
+		require.False(t, shouldMarkCreditsExhausted(resp, body, nil))
+	})
+
+	t.Run("结构化限流不标记", func(t *testing.T) {
+		resp := &http.Response{StatusCode: http.StatusTooManyRequests}
+		body := []byte(`{"error":{"status":"RESOURCE_EXHAUSTED","details":[{"@type":"type.googleapis.com/google.rpc.ErrorInfo","reason":"RATE_LIMIT_EXCEEDED"},{"@type":"type.googleapis.com/google.rpc.RetryInfo","retryDelay":"0.5s"}]}}`)
+		require.False(t, shouldMarkCreditsExhausted(resp, body, nil))
+	})
+
+	t.Run("含 credits 关键词时标记", func(t *testing.T) {
+		resp := &http.Response{StatusCode: http.StatusForbidden}
+		for _, keyword := range []string{
+			"Insufficient GOOGLE_ONE_AI credits",
+			"insufficient credit balance",
+			"not enough credits for this request",
+			"Credits exhausted",
+			"minimumCreditAmountForUsage requirement not met",
+		} {
+			body := []byte(`{"error":{"message":"` + keyword + `"}}`)
+			require.True(t, shouldMarkCreditsExhausted(resp, body, nil), "should mark for keyword: %s", keyword)
+		}
+	})
+
+	t.Run("无 credits 关键词时不标记", func(t *testing.T) {
+		resp := &http.Response{StatusCode: http.StatusForbidden}
+		body := []byte(`{"error":{"message":"permission denied"}}`)
+		require.False(t, shouldMarkCreditsExhausted(resp, body, nil))
+	})
+}
+
+func TestInjectEnabledCreditTypes(t *testing.T) {
+	t.Run("正常 JSON 注入成功", func(t *testing.T) {
+		body := []byte(`{"model":"claude-sonnet-4-5","request":{}}`)
+		result := injectEnabledCreditTypes(body)
+		require.NotNil(t, result)
+		require.Contains(t, string(result), `"enabledCreditTypes"`)
+		require.Contains(t, string(result), `GOOGLE_ONE_AI`)
+	})
+
+	t.Run("非法 JSON 返回 nil", func(t *testing.T) {
+		require.Nil(t, injectEnabledCreditTypes([]byte(`not json`)))
+	})
+
+	t.Run("空 body 返回 nil", func(t *testing.T) {
+		require.Nil(t, injectEnabledCreditTypes([]byte{}))
+	})
+
+	t.Run("已有 enabledCreditTypes 会被覆盖", func(t *testing.T) {
+		body := []byte(`{"enabledCreditTypes":["OLD"],"model":"test"}`)
+		result := injectEnabledCreditTypes(body)
+		require.NotNil(t, result)
+		require.Contains(t, string(result), `GOOGLE_ONE_AI`)
+		require.NotContains(t, string(result), `OLD`)
+	})
+}
+
+func TestClearCreditsExhausted(t *testing.T) {
+	t.Run("account 为 nil 不操作", func(t *testing.T) {
+		repo := &stubAntigravityAccountRepo{}
+		svc := &AntigravityGatewayService{accountRepo: repo}
+		svc.clearCreditsExhausted(context.Background(), nil)
+		require.Empty(t, repo.extraUpdateCalls)
+	})
+
+	t.Run("Extra 为 nil 不操作", func(t *testing.T) {
+		repo := &stubAntigravityAccountRepo{}
+		svc := &AntigravityGatewayService{accountRepo: repo}
+		svc.clearCreditsExhausted(context.Background(), &Account{ID: 1})
+		require.Empty(t, repo.extraUpdateCalls)
+	})
+
+	t.Run("无 modelRateLimitsKey 不操作", func(t *testing.T) {
+		repo := &stubAntigravityAccountRepo{}
+		svc := &AntigravityGatewayService{accountRepo: repo}
+		svc.clearCreditsExhausted(context.Background(), &Account{
+			ID:    1,
+			Extra: map[string]any{"some_key": "value"},
+		})
+		require.Empty(t, repo.extraUpdateCalls)
+	})
+
+	t.Run("无 AICredits key 不操作", func(t *testing.T) {
+		repo := &stubAntigravityAccountRepo{}
+		svc := &AntigravityGatewayService{accountRepo: repo}
+		svc.clearCreditsExhausted(context.Background(), &Account{
+			ID: 1,
+			Extra: map[string]any{
+				modelRateLimitsKey: map[string]any{
+					"claude-sonnet-4-5": map[string]any{
+						"rate_limited_at":     "2026-03-15T00:00:00Z",
+						"rate_limit_reset_at": "2099-03-15T00:00:00Z",
+					},
+				},
+			},
+		})
+		require.Empty(t, repo.extraUpdateCalls)
+	})
+
+	t.Run("有 AICredits key 时删除并调用 UpdateExtra", func(t *testing.T) {
+		repo := &stubAntigravityAccountRepo{}
+		svc := &AntigravityGatewayService{accountRepo: repo}
+		account := &Account{
+			ID: 1,
+			Extra: map[string]any{
+				modelRateLimitsKey: map[string]any{
+					"claude-sonnet-4-5": map[string]any{
+						"rate_limited_at":     "2026-03-15T00:00:00Z",
+						"rate_limit_reset_at": "2099-03-15T00:00:00Z",
+					},
+					creditsExhaustedKey: map[string]any{
+						"rate_limited_at":     "2026-03-15T00:00:00Z",
+						"rate_limit_reset_at": time.Now().Add(5 * time.Hour).UTC().Format(time.RFC3339),
+					},
+				},
+			},
+		}
+		svc.clearCreditsExhausted(context.Background(), account)
+		require.Len(t, repo.extraUpdateCalls, 1)
+		// AICredits key 应被删除
+		rawLimits := account.Extra[modelRateLimitsKey].(map[string]any)
+		_, exists := rawLimits[creditsExhaustedKey]
+		require.False(t, exists, "AICredits key 应被删除")
+		// 普通模型限流应保留
+		_, exists = rawLimits["claude-sonnet-4-5"]
+		require.True(t, exists, "普通模型限流应保留")
+	})
+}

backend/internal/service/antigravity_gateway_service.go

@@ -188,9 +188,29 @@ func (s *AntigravityGatewayService) handleSmartRetry(p antigravityRetryLoopParam
 		return &smartRetryResult{action: smartRetryActionContinueURL}
 	}
 
+	category := antigravity429Unknown
+	if resp.StatusCode == http.StatusTooManyRequests {
+		category = classifyAntigravity429(respBody)
+	}
+
 	// 判断是否触发智能重试
 	shouldSmartRetry, shouldRateLimitModel, waitDuration, modelName, isModelCapacityExhausted := shouldTriggerAntigravitySmartRetry(p.account, respBody)
 
+	// AI Credits 超量请求：
+	// 仅在上游明确返回免费配额耗尽时才允许切换到 credits。
+	if resp.StatusCode == http.StatusTooManyRequests &&
+		category == antigravity429QuotaExhausted &&
+		p.account.IsOveragesEnabled() &&
+		!p.account.isCreditsExhausted() {
+		result := s.attemptCreditsOveragesRetry(p, baseURL, modelName, waitDuration, resp.StatusCode, respBody)
+		if result.handled && result.resp != nil {
+			return &smartRetryResult{
+				action: smartRetryActionBreakWithResp,
+				resp:   result.resp,
+			}
+		}
+	}
+
 	// 情况1: retryDelay >= 阈值，限流模型并切换账号
 	if shouldRateLimitModel {
 		// 单账号 503 退避重试模式：不设限流、不切换账号，改为原地等待+重试
@@ -532,14 +552,31 @@ func (s *AntigravityGatewayService) handleSingleAccountRetryInPlace(
 
 // antigravityRetryLoop 执行带 URL fallback 的重试循环
 func (s *AntigravityGatewayService) antigravityRetryLoop(p antigravityRetryLoopParams) (*antigravityRetryLoopResult, error) {
+	// 预检查：模型限流 + overages 启用 + 积分未耗尽 → 直接注入 AI Credits
+	overagesInjected := false
+	if p.requestedModel != "" && p.account.Platform == PlatformAntigravity &&
+		p.account.IsOveragesEnabled() && !p.account.isCreditsExhausted() &&
+		p.account.isModelRateLimitedWithContext(p.ctx, p.requestedModel) {
+		if creditsBody := injectEnabledCreditTypes(p.body); creditsBody != nil {
+			p.body = creditsBody
+			overagesInjected = true
+			logger.LegacyPrintf("service.antigravity_gateway", "%s pre_check: model_rate_limited_credits_inject model=%s account=%d (injecting enabledCreditTypes)",
+				p.prefix, p.requestedModel, p.account.ID)
+		}
+	}
+
 	// 预检查：如果账号已限流，直接返回切换信号
 	if p.requestedModel != "" {
 		if remaining := p.account.GetRateLimitRemainingTimeWithContext(p.ctx, p.requestedModel); remaining > 0 {
-			// 单账号 503 退避重试模式：跳过限流预检查，直接发请求。
-			// 首次请求设的限流是为了多账号调度器跳过该账号，在单账号模式下无意义。
-			// 如果上游确实还不可用，handleSmartRetry → handleSingleAccountRetryInPlace
-			// 会在 Service 层原地等待+重试，不需要在预检查这里等。
-			if isSingleAccountRetry(p.ctx) {
+			// 已注入积分的请求不再受普通模型限流预检查阻断。
+			if overagesInjected {
+				logger.LegacyPrintf("service.antigravity_gateway", "%s pre_check: credits_injected_ignore_rate_limit remaining=%v model=%s account=%d",
+					p.prefix, remaining.Truncate(time.Millisecond), p.requestedModel, p.account.ID)
+			} else if isSingleAccountRetry(p.ctx) {
+				// 单账号 503 退避重试模式：跳过限流预检查，直接发请求。
+				// 首次请求设的限流是为了多账号调度器跳过该账号，在单账号模式下无意义。
+				// 如果上游确实还不可用，handleSmartRetry → handleSingleAccountRetryInPlace
+				// 会在 Service 层原地等待+重试，不需要在预检查这里等。
 				logger.LegacyPrintf("service.antigravity_gateway", "%s pre_check: single_account_retry skipping rate_limit remaining=%v model=%s account=%d (will retry in-place if 503)",
 					p.prefix, remaining.Truncate(time.Millisecond), p.requestedModel, p.account.ID)
 			} else {
@@ -631,6 +668,15 @@ urlFallbackLoop:
 				respBody, _ := io.ReadAll(io.LimitReader(resp.Body, 2<<20))
 				_ = resp.Body.Close()
 
+				if overagesInjected && shouldMarkCreditsExhausted(resp, respBody, nil) {
+					modelKey := resolveCreditsOveragesModelKey(p.ctx, p.account, "", p.requestedModel)
+					s.handleCreditsRetryFailure(p.ctx, p.prefix, modelKey, p.account, &http.Response{
+						StatusCode: resp.StatusCode,
+						Header:     resp.Header.Clone(),
+						Body:       io.NopCloser(bytes.NewReader(respBody)),
+					}, nil)
+				}
+
 				// ★ 统一入口：自定义错误码 + 临时不可调度
 				if handled, outStatus, policyErr := s.applyErrorPolicy(p, resp.StatusCode, resp.Header, respBody); handled {
 					if policyErr != nil {

backend/internal/service/antigravity_quota_fetcher.go

@@ -78,11 +78,11 @@ func (f *AntigravityQuotaFetcher) FetchQuota(ctx context.Context, account *Accou
 		return nil, err
 	}
 
-	// 调用 LoadCodeAssist 获取订阅等级（非关键路径，失败不影响主流程）
-	tierRaw, tierNormalized := f.fetchSubscriptionTier(ctx, client, accessToken)
+	// 调用 LoadCodeAssist 获取订阅等级和 AI Credits 余额（非关键路径，失败不影响主流程）
+	tierRaw, tierNormalized, loadResp := f.fetchSubscriptionTier(ctx, client, accessToken)
 
 	// 转换为 UsageInfo
-	usageInfo := f.buildUsageInfo(modelsResp, tierRaw, tierNormalized)
+	usageInfo := f.buildUsageInfo(modelsResp, tierRaw, tierNormalized, loadResp)
 
 	return &QuotaResult{
 		UsageInfo: usageInfo,
@@ -90,20 +90,21 @@ func (f *AntigravityQuotaFetcher) FetchQuota(ctx context.Context, account *Accou
 	}, nil
 }
 
-// fetchSubscriptionTier 获取账号订阅等级，失败返回空字符串
-func (f *AntigravityQuotaFetcher) fetchSubscriptionTier(ctx context.Context, client *antigravity.Client, accessToken string) (raw, normalized string) {
+// fetchSubscriptionTier 获取账号订阅等级，失败返回空字符串。
+// 同时返回 LoadCodeAssistResponse，以便提取 AI Credits 余额。
+func (f *AntigravityQuotaFetcher) fetchSubscriptionTier(ctx context.Context, client *antigravity.Client, accessToken string) (raw, normalized string, loadResp *antigravity.LoadCodeAssistResponse) {
 	loadResp, _, err := client.LoadCodeAssist(ctx, accessToken)
 	if err != nil {
 		slog.Warn("failed to fetch subscription tier", "error", err)
-		return "", ""
+		return "", "", nil
 	}
 	if loadResp == nil {
-		return "", ""
+		return "", "", nil
 	}
 
 	raw = loadResp.GetTier() // 已有方法：paidTier > currentTier
 	normalized = normalizeTier(raw)
-	return raw, normalized
+	return raw, normalized, loadResp
 }
 
 // normalizeTier 将原始 tier 字符串归一化为 FREE/PRO/ULTRA/UNKNOWN
@@ -124,8 +125,8 @@ func normalizeTier(raw string) string {
 	}
 }
 
-// buildUsageInfo 将 API 响应转换为 UsageInfo
-func (f *AntigravityQuotaFetcher) buildUsageInfo(modelsResp *antigravity.FetchAvailableModelsResponse, tierRaw, tierNormalized string) *UsageInfo {
+// buildUsageInfo 将 API 响应转换为 UsageInfo。
+func (f *AntigravityQuotaFetcher) buildUsageInfo(modelsResp *antigravity.FetchAvailableModelsResponse, tierRaw, tierNormalized string, loadResp *antigravity.LoadCodeAssistResponse) *UsageInfo {
 	now := time.Now()
 	info := &UsageInfo{
 		UpdatedAt:               &now,
@@ -190,6 +191,16 @@ func (f *AntigravityQuotaFetcher) buildUsageInfo(modelsResp *antigravity.FetchAv
 		}
 	}
 
+	if loadResp != nil {
+		for _, credit := range loadResp.GetAvailableCredits() {
+			info.AICredits = append(info.AICredits, AICredit{
+				CreditType:     credit.CreditType,
+				Amount:         credit.GetAmount(),
+				MinimumBalance: credit.GetMinimumAmount(),
+			})
+		}
+	}
+
 	return info
 }
 

backend/internal/service/antigravity_quota_fetcher_test.go

@@ -81,7 +81,7 @@ func TestBuildUsageInfo_BasicModels(t *testing.T) {
 		},
 	}
 
-	info := fetcher.buildUsageInfo(modelsResp, "g1-pro-tier", "PRO")
+	info := fetcher.buildUsageInfo(modelsResp, "g1-pro-tier", "PRO", nil)
 
 	// 基本字段
 	require.NotNil(t, info.UpdatedAt, "UpdatedAt should be set")
@@ -141,7 +141,7 @@ func TestBuildUsageInfo_DeprecatedModels(t *testing.T) {
 		},
 	}
 
-	info := fetcher.buildUsageInfo(modelsResp, "", "")
+	info := fetcher.buildUsageInfo(modelsResp, "", "", nil)
 
 	require.Len(t, info.ModelForwardingRules, 2)
 	require.Equal(t, "claude-sonnet-4-20250514", info.ModelForwardingRules["claude-3-sonnet-20240229"])
@@ -159,7 +159,7 @@ func TestBuildUsageInfo_NoDeprecatedModels(t *testing.T) {
 		},
 	}
 
-	info := fetcher.buildUsageInfo(modelsResp, "", "")
+	info := fetcher.buildUsageInfo(modelsResp, "", "", nil)
 
 	require.Nil(t, info.ModelForwardingRules, "ModelForwardingRules should be nil when no deprecated models")
 }
@@ -171,7 +171,7 @@ func TestBuildUsageInfo_EmptyModels(t *testing.T) {
 		Models: map[string]antigravity.ModelInfo{},
 	}
 
-	info := fetcher.buildUsageInfo(modelsResp, "", "")
+	info := fetcher.buildUsageInfo(modelsResp, "", "", nil)
 
 	require.NotNil(t, info)
 	require.NotNil(t, info.AntigravityQuota)
@@ -193,7 +193,7 @@ func TestBuildUsageInfo_ModelWithNilQuotaInfo(t *testing.T) {
 		},
 	}
 
-	info := fetcher.buildUsageInfo(modelsResp, "", "")
+	info := fetcher.buildUsageInfo(modelsResp, "", "", nil)
 
 	require.NotNil(t, info)
 	require.Empty(t, info.AntigravityQuota, "models with nil QuotaInfo should be skipped")
@@ -222,7 +222,7 @@ func TestBuildUsageInfo_FiveHourPriorityOrder(t *testing.T) {
 		},
 	}
 
-	info := fetcher.buildUsageInfo(modelsResp, "", "")
+	info := fetcher.buildUsageInfo(modelsResp, "", "", nil)
 
 	require.NotNil(t, info.FiveHour, "FiveHour should be set when a priority model exists")
 	// claude-sonnet-4-20250514 is first in priority list, so it should be used
@@ -251,7 +251,7 @@ func TestBuildUsageInfo_FiveHourFallbackToClaude4(t *testing.T) {
 		},
 	}
 
-	info := fetcher.buildUsageInfo(modelsResp, "", "")
+	info := fetcher.buildUsageInfo(modelsResp, "", "", nil)
 
 	require.NotNil(t, info.FiveHour)
 	expectedUtilization := (1.0 - 0.60) * 100 // 40
@@ -277,7 +277,7 @@ func TestBuildUsageInfo_FiveHourFallbackToGemini(t *testing.T) {
 		},
 	}
 
-	info := fetcher.buildUsageInfo(modelsResp, "", "")
+	info := fetcher.buildUsageInfo(modelsResp, "", "", nil)
 
 	require.NotNil(t, info.FiveHour)
 	expectedUtilization := (1.0 - 0.30) * 100 // 70
@@ -298,7 +298,7 @@ func TestBuildUsageInfo_FiveHourNoPriorityModel(t *testing.T) {
 		},
 	}
 
-	info := fetcher.buildUsageInfo(modelsResp, "", "")
+	info := fetcher.buildUsageInfo(modelsResp, "", "", nil)
 
 	require.Nil(t, info.FiveHour, "FiveHour should be nil when no priority model exists")
 }
@@ -317,7 +317,7 @@ func TestBuildUsageInfo_FiveHourWithEmptyResetTime(t *testing.T) {
 		},
 	}
 
-	info := fetcher.buildUsageInfo(modelsResp, "", "")
+	info := fetcher.buildUsageInfo(modelsResp, "", "", nil)
 
 	require.NotNil(t, info.FiveHour)
 	require.Nil(t, info.FiveHour.ResetsAt, "ResetsAt should be nil when ResetTime is empty")
@@ -338,7 +338,7 @@ func TestBuildUsageInfo_FullUtilization(t *testing.T) {
 		},
 	}
 
-	info := fetcher.buildUsageInfo(modelsResp, "", "")
+	info := fetcher.buildUsageInfo(modelsResp, "", "", nil)
 
 	quota := info.AntigravityQuota["claude-sonnet-4-20250514"]
 	require.NotNil(t, quota)
@@ -358,13 +358,38 @@ func TestBuildUsageInfo_ZeroUtilization(t *testing.T) {
 		},
 	}
 
-	info := fetcher.buildUsageInfo(modelsResp, "", "")
-
+	info := fetcher.buildUsageInfo(modelsResp, "", "", nil)
 	quota := info.AntigravityQuota["claude-sonnet-4-20250514"]
 	require.NotNil(t, quota)
 	require.Equal(t, 0, quota.Utilization)
 }
 
+func TestBuildUsageInfo_AICredits(t *testing.T) {
+	fetcher := &AntigravityQuotaFetcher{}
+	modelsResp := &antigravity.FetchAvailableModelsResponse{
+		Models: map[string]antigravity.ModelInfo{},
+	}
+	loadResp := &antigravity.LoadCodeAssistResponse{
+		PaidTier: &antigravity.PaidTierInfo{
+			ID: "g1-pro-tier",
+			AvailableCredits: []antigravity.AvailableCredit{
+				{
+					CreditType:                  "GOOGLE_ONE_AI",
+					CreditAmount:                "25",
+					MinimumCreditAmountForUsage: "5",
+				},
+			},
+		},
+	}
+
+	info := fetcher.buildUsageInfo(modelsResp, "g1-pro-tier", "PRO", loadResp)
+
+	require.Len(t, info.AICredits, 1)
+	require.Equal(t, "GOOGLE_ONE_AI", info.AICredits[0].CreditType)
+	require.Equal(t, 25.0, info.AICredits[0].Amount)
+	require.Equal(t, 5.0, info.AICredits[0].MinimumBalance)
+}
+
 func TestFetchQuota_ForbiddenReturnsIsForbidden(t *testing.T) {
 	// 模拟 FetchQuota 遇到 403 时的行为：
 	// FetchAvailableModels 返回 ForbiddenError → FetchQuota 应返回 is_forbidden=true

backend/internal/service/antigravity_quota_scope.go

@@ -32,6 +32,10 @@ func (a *Account) IsSchedulableForModelWithContext(ctx context.Context, requeste
 		return false
 	}
 	if a.isModelRateLimitedWithContext(ctx, requestedModel) {
+		// Antigravity + overages 启用 + 积分未耗尽 → 放行（有积分可用）
+		if a.Platform == PlatformAntigravity && a.IsOveragesEnabled() && !a.isCreditsExhausted() {
+			return true
+		}
 		return false
 	}
 	return true

backend/internal/service/antigravity_rate_limit_test.go

@@ -76,10 +76,16 @@ type modelRateLimitCall struct {
 	resetAt   time.Time
 }
 
+type extraUpdateCall struct {
+	accountID int64
+	updates   map[string]any
+}
+
 type stubAntigravityAccountRepo struct {
 	AccountRepository
 	rateCalls           []rateLimitCall
 	modelRateLimitCalls []modelRateLimitCall
+	extraUpdateCalls    []extraUpdateCall
 }
 
 func (s *stubAntigravityAccountRepo) SetRateLimited(ctx context.Context, id int64, resetAt time.Time) error {
@@ -92,6 +98,11 @@ func (s *stubAntigravityAccountRepo) SetModelRateLimit(ctx context.Context, id i
 	return nil
 }
 
+func (s *stubAntigravityAccountRepo) UpdateExtra(ctx context.Context, id int64, updates map[string]any) error {
+	s.extraUpdateCalls = append(s.extraUpdateCalls, extraUpdateCall{accountID: id, updates: updates})
+	return nil
+}
+
 func TestAntigravityRetryLoop_NoURLFallback_UsesConfiguredBaseURL(t *testing.T) {
 	t.Setenv(antigravityForwardBaseURLEnv, "")
 

backend/internal/service/antigravity_smart_retry_test.go

@@ -32,15 +32,23 @@ func (c *stubSmartRetryCache) DeleteSessionAccountID(_ context.Context, groupID
 
 // mockSmartRetryUpstream 用于 handleSmartRetry 测试的 mock upstream
 type mockSmartRetryUpstream struct {
-	responses []*http.Response
-	errors    []error
-	callIdx   int
-	calls     []string
+	responses     []*http.Response
+	errors        []error
+	callIdx       int
+	calls         []string
+	requestBodies [][]byte
 }
 
 func (m *mockSmartRetryUpstream) Do(req *http.Request, proxyURL string, accountID int64, accountConcurrency int) (*http.Response, error) {
 	idx := m.callIdx
 	m.calls = append(m.calls, req.URL.String())
+	if req != nil && req.Body != nil {
+		body, _ := io.ReadAll(req.Body)
+		m.requestBodies = append(m.requestBodies, body)
+		req.Body = io.NopCloser(bytes.NewReader(body))
+	} else {
+		m.requestBodies = append(m.requestBodies, nil)
+	}
 	m.callIdx++
 	if idx < len(m.responses) {
 		return m.responses[idx], m.errors[idx]

backend/internal/service/ratelimit_service.go

@@ -1174,7 +1174,8 @@ func hasRecoverableRuntimeState(account *Account) bool {
 	if len(account.Extra) == 0 {
 		return false
 	}
-	return hasNonEmptyMapValue(account.Extra, "model_rate_limits") || hasNonEmptyMapValue(account.Extra, "antigravity_quota_scopes")
+	return hasNonEmptyMapValue(account.Extra, "model_rate_limits") ||
+		hasNonEmptyMapValue(account.Extra, "antigravity_quota_scopes")
 }
 
 func hasNonEmptyMapValue(extra map[string]any, key string) bool {

frontend/src/components/account/AccountStatusIndicator.vue

@@ -76,19 +76,39 @@
       </div>
     </div>
 
-    <!-- Model Rate Limit Indicators (Antigravity OAuth Smart Retry) -->
+    <!-- Model Status Indicators (普通限流 / 超量请求中) -->
     <div
-      v-if="activeModelRateLimits.length > 0"
+      v-if="activeModelStatuses.length > 0"
       :class="[
-        activeModelRateLimits.length <= 4
+        activeModelStatuses.length <= 4
           ? 'flex flex-col gap-1'
-          : activeModelRateLimits.length <= 8
+          : activeModelStatuses.length <= 8
             ? 'columns-2 gap-x-2'
             : 'columns-3 gap-x-2'
       ]"
     >
-      <div v-for="item in activeModelRateLimits" :key="item.model" class="group relative mb-1 break-inside-avoid">
+      <div v-for="item in activeModelStatuses" :key="`${item.kind}-${item.model}`" class="group relative mb-1 break-inside-avoid">
+        <!-- 积分已用尽 -->
         <span
+          v-if="item.kind === 'credits_exhausted'"
+          class="inline-flex items-center gap-1 rounded bg-red-100 px-1.5 py-0.5 text-xs font-medium text-red-700 dark:bg-red-900/30 dark:text-red-400"
+        >
+          <Icon name="exclamationTriangle" size="xs" :stroke-width="2" />
+          {{ t('admin.accounts.status.creditsExhausted') }}
+          <span class="text-[10px] opacity-70">{{ formatModelResetTime(item.reset_at) }}</span>
+        </span>
+        <!-- 正在走积分（模型限流但积分可用）-->
+        <span
+          v-else-if="item.kind === 'credits_active'"
+          class="inline-flex items-center gap-1 rounded bg-amber-100 px-1.5 py-0.5 text-xs font-medium text-amber-700 dark:bg-amber-900/30 dark:text-amber-400"
+        >
+          <span>⚡</span>
+          {{ formatScopeName(item.model) }}
+          <span class="text-[10px] opacity-70">{{ formatModelResetTime(item.reset_at) }}</span>
+        </span>
+        <!-- 普通模型限流 -->
+        <span
+          v-else
           class="inline-flex items-center gap-1 rounded bg-purple-100 px-1.5 py-0.5 text-xs font-medium text-purple-700 dark:bg-purple-900/30 dark:text-purple-400"
         >
           <Icon name="exclamationTriangle" size="xs" :stroke-width="2" />
@@ -99,7 +119,13 @@
         <div
           class="pointer-events-none absolute bottom-full left-1/2 z-50 mb-2 w-56 -translate-x-1/2 whitespace-normal rounded bg-gray-900 px-3 py-2 text-center text-xs leading-relaxed text-white opacity-0 transition-opacity group-hover:opacity-100 dark:bg-gray-700"
         >
-          {{ t('admin.accounts.status.modelRateLimitedUntil', { model: formatScopeName(item.model), time: formatTime(item.reset_at) }) }}
+          {{
+            item.kind === 'credits_exhausted'
+              ? t('admin.accounts.status.creditsExhaustedUntil', { time: formatTime(item.reset_at) })
+              : item.kind === 'credits_active'
+                ? t('admin.accounts.status.modelCreditOveragesUntil', { model: formatScopeName(item.model), time: formatTime(item.reset_at) })
+                : t('admin.accounts.status.modelRateLimitedUntil', { model: formatScopeName(item.model), time: formatTime(item.reset_at) })
+          }}
           <div
             class="absolute left-1/2 top-full -translate-x-1/2 border-4 border-transparent border-t-gray-900 dark:border-t-gray-700"
           ></div>
@@ -131,6 +157,7 @@
 <script setup lang="ts">
 import { computed } from 'vue'
 import { useI18n } from 'vue-i18n'
+import Icon from '@/components/icons/Icon.vue'
 import type { Account } from '@/types'
 import { formatCountdown, formatDateTime, formatCountdownWithSuffix, formatTime } from '@/utils/format'
 
@@ -150,17 +177,44 @@ const isRateLimited = computed(() => {
   return new Date(props.account.rate_limit_reset_at) > new Date()
 })
 
+type AccountModelStatusItem = {
+  kind: 'rate_limit' | 'credits_exhausted' | 'credits_active'
+  model: string
+  reset_at: string
+}
 
-// Computed: active model rate limits (Antigravity OAuth Smart Retry)
-const activeModelRateLimits = computed(() => {
-  const modelLimits = (props.account.extra as Record<string, unknown> | undefined)?.model_rate_limits as
+// Computed: active model statuses (普通模型限流 + 积分耗尽 + 走积分中)
+const activeModelStatuses = computed<AccountModelStatusItem[]>(() => {
+  const extra = props.account.extra as Record<string, unknown> | undefined
+  const modelLimits = extra?.model_rate_limits as
     | Record<string, { rate_limited_at: string; rate_limit_reset_at: string }>
     | undefined
-  if (!modelLimits) return []
   const now = new Date()
-  return Object.entries(modelLimits)
-    .filter(([, info]) => new Date(info.rate_limit_reset_at) > now)
-    .map(([model, info]) => ({ model, reset_at: info.rate_limit_reset_at }))
+  const items: AccountModelStatusItem[] = []
+
+  if (!modelLimits) return items
+
+  // 检查 AICredits key 是否生效（积分是否耗尽）
+  const aiCreditsEntry = modelLimits['AICredits']
+  const hasActiveAICredits = aiCreditsEntry && new Date(aiCreditsEntry.rate_limit_reset_at) > now
+  const allowOverages = !!(extra?.allow_overages)
+
+  for (const [model, info] of Object.entries(modelLimits)) {
+    if (new Date(info.rate_limit_reset_at) <= now) continue
+
+    if (model === 'AICredits') {
+      // AICredits key → 积分已用尽
+      items.push({ kind: 'credits_exhausted', model, reset_at: info.rate_limit_reset_at })
+    } else if (allowOverages && !hasActiveAICredits) {
+      // 普通模型限流 + overages 启用 + 积分可用 → 正在走积分
+      items.push({ kind: 'credits_active', model, reset_at: info.rate_limit_reset_at })
+    } else {
+      // 普通模型限流
+      items.push({ kind: 'rate_limit', model, reset_at: info.rate_limit_reset_at })
+    }
+  }
+
+  return items
 })
 
 const formatScopeName = (scope: string): string => {
@@ -182,7 +236,7 @@ const formatScopeName = (scope: string): string => {
     'gemini-3.1-pro-high': 'G3PH',
     'gemini-3.1-pro-low': 'G3PL',
     'gemini-3-pro-image': 'G3PI',
-    'gemini-3.1-flash-image': 'GImage',
+    'gemini-3.1-flash-image': 'G31FI',
     // 其他
     'gpt-oss-120b-medium': 'GPT120',
     'tab_flash_lite_preview': 'TabFL',

frontend/src/components/account/AccountUsageCell.vue

@@ -289,6 +289,13 @@
           :resets-at="antigravityClaudeUsageFromAPI.resetTime"
           color="amber"
         />
+
+        <div v-if="aiCreditsDisplay" class="mt-1 text-[10px] text-gray-500 dark:text-gray-400">
+          💳 {{ t('admin.accounts.aiCreditsBalance') }}: {{ aiCreditsDisplay }}
+        </div>
+      </div>
+      <div v-else-if="aiCreditsDisplay" class="text-[10px] text-gray-500 dark:text-gray-400">
+        💳 {{ t('admin.accounts.aiCreditsBalance') }}: {{ aiCreditsDisplay }}
       </div>
       <div v-else class="text-xs text-gray-400">-</div>
     </template>
@@ -581,6 +588,14 @@ const antigravityClaudeUsageFromAPI = computed(() =>
   ])
 )
 
+const aiCreditsDisplay = computed(() => {
+  const credits = usageInfo.value?.ai_credits
+  if (!credits || credits.length === 0) return null
+  const total = credits.reduce((sum, credit) => sum + (credit.amount ?? 0), 0)
+  if (total <= 0) return null
+  return total.toFixed(0)
+})
+
 // Antigravity 账户类型（从 load_code_assist 响应中提取）
 const antigravityTier = computed(() => {
   const extra = props.account.extra as Record<string, unknown> | undefined

frontend/src/components/account/CreateAccountModal.vue

@@ -2449,6 +2449,33 @@
             </div>
           </div>
         </div>
+        <div v-if="form.platform === 'antigravity'" class="mt-3 flex items-center gap-2">
+          <label class="flex cursor-pointer items-center gap-2">
+            <input
+              type="checkbox"
+              v-model="allowOverages"
+              class="h-4 w-4 rounded border-gray-300 text-primary-500 focus:ring-primary-500 dark:border-dark-500"
+            />
+            <span class="text-sm font-medium text-gray-700 dark:text-gray-300">
+              {{ t('admin.accounts.allowOverages') }}
+            </span>
+          </label>
+          <div class="group relative">
+            <span
+              class="inline-flex h-4 w-4 cursor-help items-center justify-center rounded-full bg-gray-200 text-xs text-gray-500 hover:bg-gray-300 dark:bg-dark-600 dark:text-gray-400 dark:hover:bg-dark-500"
+            >
+              ?
+            </span>
+            <div
+              class="pointer-events-none absolute left-0 top-full z-[100] mt-1.5 w-72 rounded bg-gray-900 px-3 py-2 text-xs text-white opacity-0 transition-opacity group-hover:opacity-100 dark:bg-gray-700"
+            >
+              {{ t('admin.accounts.allowOveragesTooltip') }}
+              <div
+                class="absolute bottom-full left-3 border-4 border-transparent border-b-gray-900 dark:border-b-gray-700"
+              ></div>
+            </div>
+          </div>
+        </div>
 
         <!-- Group Selection - 仅标准模式显示 -->
         <GroupSelector
@@ -2991,6 +3018,7 @@ const openaiAPIKeyResponsesWebSocketV2Mode = ref<OpenAIWSMode>(OPENAI_WS_MODE_OF
 const codexCLIOnlyEnabled = ref(false)
 const anthropicPassthroughEnabled = ref(false)
 const mixedScheduling = ref(false) // For antigravity accounts: enable mixed scheduling
+const allowOverages = ref(false) // For antigravity accounts: enable AI Credits overages
 const antigravityAccountType = ref<'oauth' | 'upstream'>('oauth') // For antigravity: oauth or upstream
 const soraAccountType = ref<'oauth' | 'apikey'>('oauth') // For sora: oauth or apikey (upstream)
 const upstreamBaseUrl = ref('') // For upstream type: base URL
@@ -3017,6 +3045,13 @@ const getTempUnschedRuleKey = createStableObjectKeyResolver<TempUnschedRuleForm>
 const geminiOAuthType = ref<'code_assist' | 'google_one' | 'ai_studio'>('google_one')
 const geminiAIStudioOAuthEnabled = ref(false)
 
+function buildAntigravityExtra(): Record<string, unknown> | undefined {
+  const extra: Record<string, unknown> = {}
+  if (mixedScheduling.value) extra.mixed_scheduling = true
+  if (allowOverages.value) extra.allow_overages = true
+  return Object.keys(extra).length > 0 ? extra : undefined
+}
+
 const showMixedChannelWarning = ref(false)
 const mixedChannelWarningDetails = ref<{ groupName: string; currentPlatform: string; otherPlatform: string } | null>(
   null
@@ -3282,6 +3317,7 @@ watch(
       accountCategory.value = 'oauth-based'
       antigravityAccountType.value = 'oauth'
     } else {
+      allowOverages.value = false
       antigravityWhitelistModels.value = []
       antigravityModelMappings.value = []
       antigravityModelRestrictionMode.value = 'mapping'
@@ -3712,6 +3748,7 @@ const resetForm = () => {
   sessionIdMaskingEnabled.value = false
   cacheTTLOverrideEnabled.value = false
   cacheTTLOverrideTarget.value = '5m'
+  allowOverages.value = false
   antigravityAccountType.value = 'oauth'
   upstreamBaseUrl.value = ''
   upstreamApiKey.value = ''
@@ -3960,7 +3997,7 @@ const handleSubmit = async () => {
 
     applyInterceptWarmup(credentials, interceptWarmupRequests.value, 'create')
 
-    const extra = mixedScheduling.value ? { mixed_scheduling: true } : undefined
+    const extra = buildAntigravityExtra()
     await createAccountAndFinish(form.platform, 'apikey', credentials, extra)
     return
   }
@@ -4706,7 +4743,7 @@ const handleAntigravityExchange = async (authCode: string) => {
 		if (antigravityModelMapping) {
 			credentials.model_mapping = antigravityModelMapping
 		}
-		const extra = mixedScheduling.value ? { mixed_scheduling: true } : undefined
+		const extra = buildAntigravityExtra()
 		await createAccountAndFinish('antigravity', 'oauth', credentials, extra)
   } catch (error: any) {
     antigravityOAuth.error.value = error.response?.data?.detail || t('admin.accounts.oauth.authFailed')

frontend/src/components/account/EditAccountModal.vue

@@ -1610,6 +1610,33 @@
             </div>
           </div>
         </div>
+        <div v-if="account?.platform === 'antigravity'" class="mt-3 flex items-center gap-2">
+          <label class="flex cursor-pointer items-center gap-2">
+            <input
+              type="checkbox"
+              v-model="allowOverages"
+              class="h-4 w-4 rounded border-gray-300 text-primary-500 focus:ring-primary-500 dark:border-dark-500"
+            />
+            <span class="text-sm font-medium text-gray-700 dark:text-gray-300">
+              {{ t('admin.accounts.allowOverages') }}
+            </span>
+          </label>
+          <div class="group relative">
+            <span
+              class="inline-flex h-4 w-4 cursor-help items-center justify-center rounded-full bg-gray-200 text-xs text-gray-500 hover:bg-gray-300 dark:bg-dark-600 dark:text-gray-400 dark:hover:bg-dark-500"
+            >
+              ?
+            </span>
+            <div
+              class="pointer-events-none absolute left-0 top-full z-[100] mt-1.5 w-72 rounded bg-gray-900 px-3 py-2 text-xs text-white opacity-0 transition-opacity group-hover:opacity-100 dark:bg-gray-700"
+            >
+              {{ t('admin.accounts.allowOveragesTooltip') }}
+              <div
+                class="absolute bottom-full left-3 border-4 border-transparent border-b-gray-900 dark:border-b-gray-700"
+              ></div>
+            </div>
+          </div>
+        </div>
       </div>
 
       <!-- Group Selection - 仅标准模式显示 -->
@@ -1778,6 +1805,7 @@ const customErrorCodeInput = ref<number | null>(null)
 const interceptWarmupRequests = ref(false)
 const autoPauseOnExpired = ref(false)
 const mixedScheduling = ref(false) // For antigravity accounts: enable mixed scheduling
+const allowOverages = ref(false) // For antigravity accounts: enable AI Credits overages
 const antigravityModelRestrictionMode = ref<'whitelist' | 'mapping'>('whitelist')
 const antigravityWhitelistModels = ref<string[]>([])
 const antigravityModelMappings = ref<ModelMapping[]>([])
@@ -1980,8 +2008,11 @@ watch(
       autoPauseOnExpired.value = newAccount.auto_pause_on_expired === true
 
       // Load mixed scheduling setting (only for antigravity accounts)
+      mixedScheduling.value = false
+      allowOverages.value = false
       const extra = newAccount.extra as Record<string, unknown> | undefined
       mixedScheduling.value = extra?.mixed_scheduling === true
+      allowOverages.value = extra?.allow_overages === true
 
       // Load OpenAI passthrough toggle (OpenAI OAuth/API Key)
       openaiPassthroughEnabled.value = false
@@ -2822,7 +2853,7 @@ const handleSubmit = async () => {
       updatePayload.credentials = newCredentials
     }
 
-    // For antigravity accounts, handle mixed_scheduling in extra
+    // For antigravity accounts, handle mixed_scheduling and allow_overages in extra
     if (props.account.platform === 'antigravity') {
       const currentExtra = (props.account.extra as Record<string, unknown>) || {}
       const newExtra: Record<string, unknown> = { ...currentExtra }
@@ -2831,6 +2862,11 @@ const handleSubmit = async () => {
       } else {
         delete newExtra.mixed_scheduling
       }
+      if (allowOverages.value) {
+        newExtra.allow_overages = true
+      } else {
+        delete newExtra.allow_overages
+      }
       updatePayload.extra = newExtra
     }
 

frontend/src/components/account/__tests__/AccountStatusIndicator.spec.ts

@@ -0,0 +1,162 @@
+import { describe, expect, it, vi } from 'vitest'
+import { mount } from '@vue/test-utils'
+import AccountStatusIndicator from '../AccountStatusIndicator.vue'
+import type { Account } from '@/types'
+
+vi.mock('vue-i18n', async () => {
+  const actual = await vi.importActual<typeof import('vue-i18n')>('vue-i18n')
+  return {
+    ...actual,
+    useI18n: () => ({
+      t: (key: string) => key
+    })
+  }
+})
+
+function makeAccount(overrides: Partial<Account>): Account {
+  return {
+    id: 1,
+    name: 'account',
+    platform: 'antigravity',
+    type: 'oauth',
+    proxy_id: null,
+    concurrency: 1,
+    priority: 1,
+    status: 'active',
+    error_message: null,
+    last_used_at: null,
+    expires_at: null,
+    auto_pause_on_expired: true,
+    created_at: '2026-03-15T00:00:00Z',
+    updated_at: '2026-03-15T00:00:00Z',
+    schedulable: true,
+    rate_limited_at: null,
+    rate_limit_reset_at: null,
+    overload_until: null,
+    temp_unschedulable_until: null,
+    temp_unschedulable_reason: null,
+    session_window_start: null,
+    session_window_end: null,
+    session_window_status: null,
+    ...overrides,
+  }
+}
+
+describe('AccountStatusIndicator', () => {
+  it('模型限流 + overages 启用 + 无 AICredits key → 显示 ⚡ (credits_active)', () => {
+    const wrapper = mount(AccountStatusIndicator, {
+      props: {
+        account: makeAccount({
+          id: 1,
+          name: 'ag-1',
+          extra: {
+            allow_overages: true,
+            model_rate_limits: {
+              'claude-sonnet-4-5': {
+                rate_limited_at: '2026-03-15T00:00:00Z',
+                rate_limit_reset_at: '2099-03-15T00:00:00Z'
+              }
+            }
+          }
+        })
+      },
+      global: {
+        stubs: {
+          Icon: true
+        }
+      }
+    })
+
+    expect(wrapper.text()).toContain('⚡')
+    expect(wrapper.text()).toContain('CSon45')
+  })
+
+  it('模型限流 + overages 未启用 → 普通限流样式（无 ⚡）', () => {
+    const wrapper = mount(AccountStatusIndicator, {
+      props: {
+        account: makeAccount({
+          id: 2,
+          name: 'ag-2',
+          extra: {
+            model_rate_limits: {
+              'claude-sonnet-4-5': {
+                rate_limited_at: '2026-03-15T00:00:00Z',
+                rate_limit_reset_at: '2099-03-15T00:00:00Z'
+              }
+            }
+          }
+        })
+      },
+      global: {
+        stubs: {
+          Icon: true
+        }
+      }
+    })
+
+    expect(wrapper.text()).toContain('CSon45')
+    expect(wrapper.text()).not.toContain('⚡')
+  })
+
+  it('AICredits key 生效 → 显示积分已用尽 (credits_exhausted)', () => {
+    const wrapper = mount(AccountStatusIndicator, {
+      props: {
+        account: makeAccount({
+          id: 3,
+          name: 'ag-3',
+          extra: {
+            allow_overages: true,
+            model_rate_limits: {
+              'AICredits': {
+                rate_limited_at: '2026-03-15T00:00:00Z',
+                rate_limit_reset_at: '2099-03-15T00:00:00Z'
+              }
+            }
+          }
+        })
+      },
+      global: {
+        stubs: {
+          Icon: true
+        }
+      }
+    })
+
+    expect(wrapper.text()).toContain('account.creditsExhausted')
+  })
+
+  it('模型限流 + overages 启用 + AICredits key 生效 → 普通限流样式（积分耗尽，无 ⚡）', () => {
+    const wrapper = mount(AccountStatusIndicator, {
+      props: {
+        account: makeAccount({
+          id: 4,
+          name: 'ag-4',
+          extra: {
+            allow_overages: true,
+            model_rate_limits: {
+              'claude-sonnet-4-5': {
+                rate_limited_at: '2026-03-15T00:00:00Z',
+                rate_limit_reset_at: '2099-03-15T00:00:00Z'
+              },
+              'AICredits': {
+                rate_limited_at: '2026-03-15T00:00:00Z',
+                rate_limit_reset_at: '2099-03-15T00:00:00Z'
+              }
+            }
+          }
+        })
+      },
+      global: {
+        stubs: {
+          Icon: true
+        }
+      }
+    })
+
+    // 模型限流 + 积分耗尽 → 不应显示 ⚡
+    expect(wrapper.text()).toContain('CSon45')
+    expect(wrapper.text()).not.toContain('⚡')
+    // AICredits 积分耗尽状态应显示
+    expect(wrapper.text()).toContain('account.creditsExhausted')
+  })
+})

frontend/src/components/account/__tests__/AccountUsageCell.spec.ts

@@ -1,6 +1,7 @@
 import { describe, expect, it, vi, beforeEach } from 'vitest'
 import { flushPromises, mount } from '@vue/test-utils'
 import AccountUsageCell from '../AccountUsageCell.vue'
+import type { Account } from '@/types'
 
 const { getUsage } = vi.hoisted(() => ({
   getUsage: vi.fn()
@@ -24,6 +25,35 @@ vi.mock('vue-i18n', async () => {
   }
 })
 
+function makeAccount(overrides: Partial<Account>): Account {
+  return {
+    id: 1,
+    name: 'account',
+    platform: 'antigravity',
+    type: 'oauth',
+    proxy_id: null,
+    concurrency: 1,
+    priority: 1,
+    status: 'active',
+    error_message: null,
+    last_used_at: null,
+    expires_at: null,
+    auto_pause_on_expired: true,
+    created_at: '2026-03-15T00:00:00Z',
+    updated_at: '2026-03-15T00:00:00Z',
+    schedulable: true,
+    rate_limited_at: null,
+    rate_limit_reset_at: null,
+    overload_until: null,
+    temp_unschedulable_until: null,
+    temp_unschedulable_reason: null,
+    session_window_start: null,
+    session_window_end: null,
+    session_window_status: null,
+    ...overrides,
+  }
+}
+
 describe('AccountUsageCell', () => {
   beforeEach(() => {
     getUsage.mockReset()
@@ -49,12 +79,12 @@ describe('AccountUsageCell', () => {
 
     const wrapper = mount(AccountUsageCell, {
       props: {
-        account: {
+        account: makeAccount({
           id: 1001,
           platform: 'antigravity',
           type: 'oauth',
           extra: {}
-        } as any
+        })
       },
       global: {
         stubs: {
@@ -72,6 +102,40 @@ describe('AccountUsageCell', () => {
     expect(wrapper.text()).toContain('admin.accounts.usageWindow.gemini3Image|70|2026-03-01T09:00:00Z')
   })
 
+  it('Antigravity 会显示 AI Credits 余额信息', async () => {
+    getUsage.mockResolvedValue({
+      ai_credits: [
+        {
+          credit_type: 'GOOGLE_ONE_AI',
+          amount: 25,
+          minimum_balance: 5
+        }
+      ]
+    })
+
+    const wrapper = mount(AccountUsageCell, {
+      props: {
+        account: makeAccount({
+          id: 1002,
+          platform: 'antigravity',
+          type: 'oauth',
+          extra: {}
+        })
+      },
+      global: {
+        stubs: {
+          UsageProgressBar: true,
+          AccountQuotaInfo: true
+        }
+      }
+    })
+
+    await flushPromises()
+
+    expect(wrapper.text()).toContain('admin.accounts.aiCreditsBalance')
+    expect(wrapper.text()).toContain('25')
+  })
+
 
   it('OpenAI OAuth 快照已过期时首屏会重新请求 usage', async () => {
     getUsage.mockResolvedValue({
@@ -103,7 +167,7 @@ describe('AccountUsageCell', () => {
 
     const wrapper = mount(AccountUsageCell, {
       props: {
-        account: {
+        account: makeAccount({
           id: 2000,
           platform: 'openai',
           type: 'oauth',
@@ -114,7 +178,7 @@ describe('AccountUsageCell', () => {
             codex_7d_used_percent: 34,
             codex_7d_reset_at: '2026-03-13T12:00:00Z'
           }
-        } as any
+        })
       },
       global: {
         stubs: {
@@ -137,7 +201,7 @@ describe('AccountUsageCell', () => {
   it('OpenAI OAuth 有现成快照且未限额时不会首屏请求 usage', async () => {
     const wrapper = mount(AccountUsageCell, {
       props: {
-        account: {
+        account: makeAccount({
           id: 2001,
           platform: 'openai',
           type: 'oauth',
@@ -148,7 +212,7 @@ describe('AccountUsageCell', () => {
             codex_7d_used_percent: 34,
             codex_7d_reset_at: '2099-03-13T12:00:00Z'
           }
-        } as any
+        })
       },
       global: {
         stubs: {
@@ -196,15 +260,15 @@ describe('AccountUsageCell', () => {
 	  }
 	})
 
-	const wrapper = mount(AccountUsageCell, {
-	  props: {
-	    account: {
-	      id: 2002,
-	      platform: 'openai',
-	      type: 'oauth',
-	      extra: {}
-	    } as any
-	  },
+		const wrapper = mount(AccountUsageCell, {
+		  props: {
+		    account: makeAccount({
+		      id: 2002,
+		      platform: 'openai',
+		      type: 'oauth',
+		      extra: {}
+		    })
+		  },
 	  global: {
 	    stubs: {
 	      UsageProgressBar: {
@@ -256,16 +320,16 @@ describe('AccountUsageCell', () => {
 	    seven_day: null
 	  })
 
-	const wrapper = mount(AccountUsageCell, {
-	  props: {
-	    account: {
-	      id: 2003,
-	      platform: 'openai',
-	      type: 'oauth',
-	      updated_at: '2026-03-07T10:00:00Z',
-	      extra: {}
-	    } as any
-	  },
+		const wrapper = mount(AccountUsageCell, {
+		  props: {
+		    account: makeAccount({
+		      id: 2003,
+		      platform: 'openai',
+		      type: 'oauth',
+		      updated_at: '2026-03-07T10:00:00Z',
+		      extra: {}
+		    })
+		  },
 	  global: {
 	    stubs: {
 	      UsageProgressBar: {
@@ -324,19 +388,19 @@ describe('AccountUsageCell', () => {
 	  }
 	})
 
-	const wrapper = mount(AccountUsageCell, {
-	  props: {
-	    account: {
-	      id: 2004,
-	      platform: 'openai',
-	      type: 'oauth',
-	      rate_limit_reset_at: '2099-03-07T12:00:00Z',
-	      extra: {
-	        codex_5h_used_percent: 0,
-	        codex_7d_used_percent: 0
-	      }
-	    } as any
-	  },
+		const wrapper = mount(AccountUsageCell, {
+		  props: {
+		    account: makeAccount({
+		      id: 2004,
+		      platform: 'openai',
+		      type: 'oauth',
+		      rate_limit_reset_at: '2099-03-07T12:00:00Z',
+		      extra: {
+		        codex_5h_used_percent: 0,
+		        codex_7d_used_percent: 0
+		      }
+		    })
+		  },
 	  global: {
 	    stubs: {
 	      UsageProgressBar: {

frontend/src/i18n/locales/en.ts

@@ -245,6 +245,7 @@ export default {
   // Common
   common: {
     loading: 'Loading...',
+    justNow: 'just now',
     save: 'Save',
     cancel: 'Cancel',
     delete: 'Delete',
@@ -1655,6 +1656,14 @@ export default {
         enabled: 'Enabled',
         disabled: 'Disabled'
       },
+      claudeMaxSimulation: {
+        title: 'Claude Max Usage Simulation',
+        tooltip:
+          'When enabled, for Claude models without upstream cache-write usage, the system deterministically maps tokens to a small input plus 1h cache creation while keeping total tokens unchanged.',
+        enabled: 'Enabled (simulate 1h cache)',
+        disabled: 'Disabled',
+        hint: 'Only token categories in usage billing logs are adjusted. No per-request mapping state is persisted.'
+      },
       supportedScopes: {
         title: 'Supported Model Families',
         tooltip: 'Select the model families this group supports. Unchecked families will not be routed to this group.',
@@ -1867,6 +1876,9 @@ export default {
         rateLimitedUntil: 'Rate limited and removed from scheduling. Auto resumes at {time}',
         rateLimitedAutoResume: 'Auto resumes in {time}',
         modelRateLimitedUntil: '{model} rate limited until {time}',
+        modelCreditOveragesUntil: '{model} using AI Credits until {time}',
+        creditsExhausted: 'Credits Exhausted',
+        creditsExhaustedUntil: 'AI Credits exhausted, expected recovery at {time}',
         overloadedUntil: 'Overloaded until {time}',
         viewTempUnschedDetails: 'View temp unschedulable details'
       },
@@ -1968,7 +1980,7 @@ export default {
       resetQuota: 'Reset Quota',
       quotaLimit: 'Quota Limit',
       quotaLimitPlaceholder: '0 means unlimited',
-      quotaLimitHint: 'Set daily/weekly/total spending limits (USD). Account will be paused when any limit is reached. Changing limits won\'t reset usage.',
+      quotaLimitHint: 'Set daily/weekly/total spending limits (USD). Anthropic API key accounts can also configure client affinity. Changing limits won\'t reset usage.',
       quotaLimitToggle: 'Enable Quota Limit',
       quotaLimitToggleHint: 'When enabled, account will be paused when usage reaches the set limit',
       quotaDailyLimit: 'Daily Limit',
@@ -2165,7 +2177,7 @@ export default {
       // Quota control (Anthropic OAuth/SetupToken only)
       quotaControl: {
         title: 'Quota Control',
-        hint: 'Only applies to Anthropic OAuth/Setup Token accounts',
+        hint: 'Configure cost window, session limits, client affinity and other scheduling controls.',
         windowCost: {
           label: '5h Window Cost Limit',
           hint: 'Limit account cost usage within the 5-hour window',
@@ -2220,8 +2232,26 @@ export default {
           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'
+        },
+        clientAffinity: {
+          label: 'Client Affinity Scheduling',
+          hint: 'When enabled, new sessions prefer accounts previously used by this client to reduce account switching'
         }
       },
+      affinityNoClients: 'No affinity clients',
+      affinityClients: '{count} affinity clients:',
+      affinitySection: 'Client Affinity',
+      affinitySectionHint: 'Control how clients are distributed across accounts. Configure zone thresholds to balance load.',
+      affinityToggle: 'Enable Client Affinity',
+      affinityToggleHint: 'New sessions prefer accounts previously used by this client',
+      affinityBase: 'Base Limit (Green Zone)',
+      affinityBasePlaceholder: 'Empty = no limit',
+      affinityBaseHint: 'Max clients in green zone (full priority scheduling)',
+      affinityBaseOffHint: 'No green zone limit. All clients receive full priority scheduling.',
+      affinityBuffer: 'Buffer (Yellow Zone)',
+      affinityBufferPlaceholder: 'e.g. 3',
+      affinityBufferHint: 'Additional clients allowed in the yellow zone (degraded priority)',
+      affinityBufferInfinite: 'Unlimited',
       expired: 'Expired',
       proxy: 'Proxy',
       noProxy: 'No Proxy',
@@ -2239,6 +2269,10 @@ export default {
       mixedSchedulingHint: 'Enable to participate in Anthropic/Gemini group scheduling',
       mixedSchedulingTooltip:
         '!! WARNING !! Antigravity Claude and Anthropic Claude cannot be used in the same context. If you have both Anthropic and Antigravity accounts, enabling this option will cause frequent 400 errors. When enabled, please use the group feature to isolate Antigravity accounts from Anthropic accounts. Make sure you understand this before enabling!!',
+      aiCreditsBalance: 'AI Credits',
+      allowOverages: 'Allow Overages (AI Credits)',
+      allowOveragesTooltip:
+        'Only use AI Credits after free quota is explicitly exhausted. Ordinary concurrent 429 rate limits will not switch to overages.',
       creating: 'Creating...',
       updating: 'Updating...',
       accountCreated: 'Account created successfully',
@@ -2672,7 +2706,7 @@ export default {
         geminiFlashDaily: 'Flash',
         gemini3Pro: 'G3P',
         gemini3Flash: 'G3F',
-        gemini3Image: 'GImage',
+        gemini3Image: 'G31FI',
         claude: 'Claude'
       },
       tier: {
@@ -4185,40 +4219,55 @@ export default {
         usage: 'Usage: Add to request header - x-api-key: <your-admin-api-key>'
       },
       soraS3: {
-        title: 'Sora S3 Storage',
-        description: 'Manage multiple Sora S3 endpoints and switch the active profile',
+        title: 'Sora Storage',
+        description: 'Manage Sora media storage profiles with S3 and Google Drive support',
         newProfile: 'New Profile',
         reloadProfiles: 'Reload Profiles',
-        empty: 'No Sora S3 profiles yet, create one first',
-        createTitle: 'Create Sora S3 Profile',
-        editTitle: 'Edit Sora S3 Profile',
+        empty: 'No storage profiles yet, create one first',
+        createTitle: 'Create Storage Profile',
+        editTitle: 'Edit Storage Profile',
+        selectProvider: 'Select Storage Type',
+        providerS3Desc: 'S3-compatible object storage',
+        providerGDriveDesc: 'Google Drive cloud storage',
         profileID: 'Profile ID',
         profileName: 'Profile Name',
         setActive: 'Set as active after creation',
         saveProfile: 'Save Profile',
         activateProfile: 'Activate',
-        profileCreated: 'Sora S3 profile created',
-        profileSaved: 'Sora S3 profile saved',
-        profileDeleted: 'Sora S3 profile deleted',
-        profileActivated: 'Sora S3 active profile switched',
+        profileCreated: 'Storage profile created',
+        profileSaved: 'Storage profile saved',
+        profileDeleted: 'Storage profile deleted',
+        profileActivated: 'Active storage profile switched',
         profileIDRequired: 'Profile ID is required',
         profileNameRequired: 'Profile name is required',
         profileSelectRequired: 'Please select a profile first',
         endpointRequired: 'S3 endpoint is required when enabled',
         bucketRequired: 'Bucket is required when enabled',
         accessKeyRequired: 'Access Key ID is required when enabled',
-        deleteConfirm: 'Delete Sora S3 profile {profileID}?',
+        deleteConfirm: 'Delete storage profile {profileID}?',
         columns: {
           profile: 'Profile',
+          profileId: 'Profile ID',
+          name: 'Name',
+          provider: 'Type',
           active: 'Active',
           endpoint: 'Endpoint',
-          bucket: 'Bucket',
+          storagePath: 'Storage Path',
+          capacityUsage: 'Capacity / Used',
+          capacityUnlimited: 'Unlimited',
+          videoCount: 'Videos',
+          videoCompleted: 'completed',
+          videoInProgress: 'in progress',
           quota: 'Default Quota',
           updatedAt: 'Updated At',
-          actions: 'Actions'
+          actions: 'Actions',
+          rootFolder: 'Root folder',
+          testInTable: 'Test',
+          testingInTable: 'Testing...',
+          testTimeout: 'Test timed out (15s)'
         },
-        enabled: 'Enable S3 Storage',
-        enabledHint: 'When enabled, Sora generated media files will be automatically uploaded to S3 storage',
+        enabled: 'Enable Storage',
+        enabledHint: 'When enabled, Sora generated media files will be automatically uploaded',
         endpoint: 'S3 Endpoint',
         region: 'Region',
         bucket: 'Bucket',
@@ -4227,16 +4276,38 @@ export default {
         secretAccessKey: 'Secret Access Key',
         secretConfigured: '(Configured, leave blank to keep)',
         cdnUrl: 'CDN URL',
-        cdnUrlHint: 'Optional. When configured, files are accessed via CDN URL instead of presigned URLs',
+        cdnUrlHint: 'Optional. When configured, files are accessed via CDN URL',
         forcePathStyle: 'Force Path Style',
         defaultQuota: 'Default Storage Quota',
         defaultQuotaHint: 'Default quota when not specified at user or group level. 0 means unlimited',
         testConnection: 'Test Connection',
         testing: 'Testing...',
-        testSuccess: 'S3 connection test successful',
-        testFailed: 'S3 connection test failed',
-        saved: 'Sora S3 settings saved successfully',
-        saveFailed: 'Failed to save Sora S3 settings'
+        testSuccess: 'Connection test successful',
+        testFailed: 'Connection test failed',
+        saved: 'Storage settings saved successfully',
+        saveFailed: 'Failed to save storage settings',
+        gdrive: {
+          authType: 'Authentication Method',
+          serviceAccount: 'Service Account',
+          clientId: 'Client ID',
+          clientSecret: 'Client Secret',
+          clientSecretConfigured: '(Configured, leave blank to keep)',
+          refreshToken: 'Refresh Token',
+          refreshTokenConfigured: '(Configured, leave blank to keep)',
+          serviceAccountJson: 'Service Account JSON',
+          serviceAccountConfigured: '(Configured, leave blank to keep)',
+          folderId: 'Folder ID (optional)',
+          authorize: 'Authorize Google Drive',
+          authorizeHint: 'Get Refresh Token via OAuth2',
+          oauthFieldsRequired: 'Please fill in Client ID and Client Secret first',
+          oauthSuccess: 'Google Drive authorization successful',
+          oauthFailed: 'Google Drive authorization failed',
+          closeWindow: 'This window will close automatically',
+          processing: 'Processing authorization...',
+          testStorage: 'Test Storage',
+          testSuccess: 'Google Drive storage test passed (upload, access, delete all OK)',
+          testFailed: 'Google Drive storage test failed'
+        }
       },
       streamTimeout: {
         title: 'Stream Timeout Handling',
@@ -4707,6 +4778,7 @@ export default {
     downloadLocal: 'Download',
     canDownload: 'to download',
     regenrate: 'Regenerate',
+    regenerate: 'Regenerate',
     creatorPlaceholder: 'Describe the video or image you want to create...',
     videoModels: 'Video Models',
     imageModels: 'Image Models',
@@ -4723,6 +4795,13 @@ export default {
     galleryEmptyTitle: 'No works yet',
     galleryEmptyDesc: 'Your creations will be displayed here. Go to the generate page to start your first creation.',
     startCreating: 'Start Creating',
-    yesterday: 'Yesterday'
+    yesterday: 'Yesterday',
+    landscape: 'Landscape',
+    portrait: 'Portrait',
+    square: 'Square',
+    examplePrompt1: 'A golden Shiba Inu walking through the streets of Shibuya, Tokyo, camera following, cinematic shot, 4K',
+    examplePrompt2: 'Drone aerial view, green aurora reflecting on a glacial lake in Iceland, slow push-in',
+    examplePrompt3: 'Cyberpunk futuristic city, neon lights reflected in rain puddles, nightscape, cinematic colors',
+    examplePrompt4: 'Chinese ink painting style, a small boat drifting among misty mountains and rivers, classical atmosphere'
   }
 }

frontend/src/i18n/locales/zh.ts

@@ -245,6 +245,7 @@ export default {
   // Common
   common: {
     loading: '加载中...',
+    justNow: '刚刚',
     save: '保存',
     cancel: '取消',
     delete: '删除',
@@ -1974,7 +1975,7 @@ export default {
       resetQuota: '重置配额',
       quotaLimit: '配额限制',
       quotaLimitPlaceholder: '0 表示不限制',
-      quotaLimitHint: '设置日/周/总使用额度（美元），任一维度达到限额后账号暂停调度。修改限额不会重置已用额度。',
+      quotaLimitHint: '设置日/周/总使用额度（美元），任一维度达到限额后账号暂停调度。Anthropic API Key 账号还可配置客户端亲和。修改限额不会重置已用额度。',
       quotaLimitToggle: '启用配额限制',
       quotaLimitToggleHint: '开启后，当账号用量达到设定额度时自动暂停调度',
       quotaDailyLimit: '日限额',
@@ -2052,6 +2053,9 @@ export default {
         rateLimitedUntil: '限流中，当前不参与调度，预计 {time} 自动恢复',
         rateLimitedAutoResume: '{time} 自动恢复',
         modelRateLimitedUntil: '{model} 限流至 {time}',
+        modelCreditOveragesUntil: '{model} 正在使用 AI Credits，至 {time}',
+        creditsExhausted: '积分已用尽',
+        creditsExhaustedUntil: 'AI Credits 已用尽，预计 {time} 恢复',
         overloadedUntil: '负载过重，重置时间：{time}',
         viewTempUnschedDetails: '查看临时不可调度详情'
       },
@@ -2105,7 +2109,7 @@ export default {
         geminiFlashDaily: 'Flash',
         gemini3Pro: 'G3P',
         gemini3Flash: 'G3F',
-        gemini3Image: 'GImage',
+        gemini3Image: 'G31FI',
         claude: 'Claude'
       },
       tier: {
@@ -2315,7 +2319,7 @@ export default {
       // Quota control (Anthropic OAuth/SetupToken only)
       quotaControl: {
         title: '配额控制',
-        hint: '仅适用于 Anthropic OAuth/Setup Token 账号',
+        hint: '配置费用窗口、会话限制、客户端亲和等调度控制。',
         windowCost: {
           label: '5h窗口费用控制',
           hint: '限制账号在5小时窗口内的费用使用',
@@ -2370,8 +2374,26 @@ export default {
           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'
+        },
+        clientAffinity: {
+          label: '客户端亲和调度',
+          hint: '启用后，新会话会优先调度到该客户端之前使用过的账号，避免频繁切换账号'
         }
       },
+      affinityNoClients: '无亲和客户端',
+      affinityClients: '{count} 个亲和客户端：',
+      affinitySection: '客户端亲和',
+      affinitySectionHint: '控制客户端在账号间的分布。通过配置区域阈值来平衡负载。',
+      affinityToggle: '启用客户端亲和',
+      affinityToggleHint: '新会话优先调度到该客户端之前使用过的账号',
+      affinityBase: '基础限额（绿区）',
+      affinityBasePlaceholder: '留空表示不限制',
+      affinityBaseHint: '绿区最大客户端数量（完整优先级调度）',
+      affinityBaseOffHint: '未开启绿区限制，所有客户端均享受完整优先级调度',
+      affinityBuffer: '缓冲区（黄区）',
+      affinityBufferPlaceholder: '例如 3',
+      affinityBufferHint: '黄区允许的额外客户端数量（降级优先级调度）',
+      affinityBufferInfinite: '不限制',
       expired: '已过期',
       proxy: '代理',
       noProxy: '无代理',
@@ -2389,6 +2411,10 @@ export default {
       mixedSchedulingHint: '启用后可参与 Anthropic/Gemini 分组的调度',
       mixedSchedulingTooltip:
         '！！注意！！ Antigravity Claude 和 Anthropic Claude 无法在同个上下文中使用，如果你同时有 Anthropic 账号和 Antigravity 账号，开启此选项会导致经常 400 报错。开启后，请用分组功能做好 Antigravity 账号和 Anthropic 账号的隔离。一定要弄明白再开启！！',
+      aiCreditsBalance: 'AI Credits',
+      allowOverages: '允许超量请求 (AI Credits)',
+      allowOveragesTooltip:
+        '仅在免费配额被明确判定为耗尽后才会使用 AI Credits。普通并发 429 限流不会切换到超量请求。',
       creating: '创建中...',
       updating: '更新中...',
       accountCreated: '账号创建成功',
@@ -4358,40 +4384,55 @@ export default {
         usage: '使用方法：在请求头中添加 x-api-key: <your-admin-api-key>'
       },
       soraS3: {
-        title: 'Sora S3 存储配置',
-        description: '以多配置列表方式管理 Sora S3 端点，并可切换生效配置',
+        title: 'Sora 存储配置',
+        description: '以多配置列表管理 Sora 媒体存储，支持 S3 和 Google Drive',
         newProfile: '新建配置',
         reloadProfiles: '刷新列表',
-        empty: '暂无 Sora S3 配置，请先创建',
-        createTitle: '新建 Sora S3 配置',
-        editTitle: '编辑 Sora S3 配置',
+        empty: '暂无存储配置，请先创建',
+        createTitle: '新建存储配置',
+        editTitle: '编辑存储配置',
+        selectProvider: '选择存储类型',
+        providerS3Desc: 'S3 兼容对象存储',
+        providerGDriveDesc: 'Google Drive 云盘',
         profileID: '配置 ID',
         profileName: '配置名称',
         setActive: '创建后设为生效',
         saveProfile: '保存配置',
         activateProfile: '设为生效',
-        profileCreated: 'Sora S3 配置创建成功',
-        profileSaved: 'Sora S3 配置保存成功',
-        profileDeleted: 'Sora S3 配置删除成功',
-        profileActivated: 'Sora S3 生效配置已切换',
+        profileCreated: '存储配置创建成功',
+        profileSaved: '存储配置保存成功',
+        profileDeleted: '存储配置删除成功',
+        profileActivated: '生效配置已切换',
         profileIDRequired: '请填写配置 ID',
         profileNameRequired: '请填写配置名称',
         profileSelectRequired: '请先选择配置',
         endpointRequired: '启用时必须填写 S3 端点',
         bucketRequired: '启用时必须填写存储桶',
         accessKeyRequired: '启用时必须填写 Access Key ID',
-        deleteConfirm: '确定删除 Sora S3 配置 {profileID} 吗？',
+        deleteConfirm: '确定删除存储配置 {profileID} 吗？',
         columns: {
           profile: '配置',
+          profileId: 'Profile ID',
+          name: '名称',
+          provider: '存储类型',
           active: '生效状态',
           endpoint: '端点',
-          bucket: '存储桶',
+          storagePath: '存储路径',
+          capacityUsage: '容量 / 已用',
+          capacityUnlimited: '无限制',
+          videoCount: '视频数',
+          videoCompleted: '完成',
+          videoInProgress: '进行中',
           quota: '默认配额',
           updatedAt: '更新时间',
-          actions: '操作'
+          actions: '操作',
+          rootFolder: '根目录',
+          testInTable: '测试',
+          testingInTable: '测试中...',
+          testTimeout: '测试超时（15秒）'
         },
-        enabled: '启用 S3 存储',
-        enabledHint: '启用后，Sora 生成的媒体文件将自动上传到 S3 存储',
+        enabled: '启用存储',
+        enabledHint: '启用后，Sora 生成的媒体文件将自动上传到存储',
         endpoint: 'S3 端点',
         region: '区域',
         bucket: '存储桶',
@@ -4400,16 +4441,38 @@ export default {
         secretAccessKey: 'Secret Access Key',
         secretConfigured: '(已配置，留空保持不变)',
         cdnUrl: 'CDN URL',
-        cdnUrlHint: '可选，配置后使用 CDN URL 访问文件，否则使用预签名 URL',
+        cdnUrlHint: '可选，配置后使用 CDN URL 访问文件',
         forcePathStyle: '强制路径风格（Path Style）',
         defaultQuota: '默认存储配额',
         defaultQuotaHint: '未在用户或分组级别指定配额时的默认值，0 表示无限制',
         testConnection: '测试连接',
         testing: '测试中...',
-        testSuccess: 'S3 连接测试成功',
-        testFailed: 'S3 连接测试失败',
-        saved: 'Sora S3 设置保存成功',
-        saveFailed: '保存 Sora S3 设置失败'
+        testSuccess: '连接测试成功',
+        testFailed: '连接测试失败',
+        saved: '存储设置保存成功',
+        saveFailed: '保存存储设置失败',
+        gdrive: {
+          authType: '认证方式',
+          serviceAccount: '服务账号',
+          clientId: 'Client ID',
+          clientSecret: 'Client Secret',
+          clientSecretConfigured: '(已配置，留空保持不变)',
+          refreshToken: 'Refresh Token',
+          refreshTokenConfigured: '(已配置，留空保持不变)',
+          serviceAccountJson: '服务账号 JSON',
+          serviceAccountConfigured: '(已配置，留空保持不变)',
+          folderId: 'Folder ID（可选）',
+          authorize: '授权 Google Drive',
+          authorizeHint: '通过 OAuth2 获取 Refresh Token',
+          oauthFieldsRequired: '请先填写 Client ID 和 Client Secret',
+          oauthSuccess: 'Google Drive 授权成功',
+          oauthFailed: 'Google Drive 授权失败',
+          closeWindow: '此窗口将自动关闭',
+          processing: '正在处理授权...',
+          testStorage: '测试存储',
+          testSuccess: 'Google Drive 存储测试成功（上传、访问、删除均正常）',
+          testFailed: 'Google Drive 存储测试失败'
+        }
       },
       streamTimeout: {
         title: '流超时处理',
@@ -4905,6 +4968,7 @@ export default {
     downloadLocal: '本地下载',
     canDownload: '可下载',
     regenrate: '重新生成',
+    regenerate: '重新生成',
     creatorPlaceholder: '描述你想要生成的视频或图片...',
     videoModels: '视频模型',
     imageModels: '图片模型',
@@ -4921,6 +4985,13 @@ export default {
     galleryEmptyTitle: '还没有任何作品',
     galleryEmptyDesc: '你的创作成果将会展示在这里。前往生成页，开始你的第一次创作吧。',
     startCreating: '开始创作',
-    yesterday: '昨天'
+    yesterday: '昨天',
+    landscape: '横屏',
+    portrait: '竖屏',
+    square: '方形',
+    examplePrompt1: '一只金色的柴犬在东京涩谷街头散步，镜头跟随，电影感画面，4K 高清',
+    examplePrompt2: '无人机航拍视角，冰岛极光下的冰川湖面反射绿色光芒，慢速推进',
+    examplePrompt3: '赛博朋克风格的未来城市，霓虹灯倒映在雨后积水中，夜景，电影级色彩',
+    examplePrompt4: '水墨画风格，一叶扁舟在山水间漂泊，薄雾缭绕，中国古典意境'
   }
 }

frontend/src/types/index.ts

@@ -403,6 +403,8 @@ export interface AdminGroup extends Group {
 
   // MCP XML 协议注入（仅 antigravity 平台使用）
   mcp_xml_inject: boolean
+  // Claude usage 模拟开关（仅 anthropic 平台使用）
+  simulate_claude_max_enabled: boolean
 
   // 支持的模型系列（仅 antigravity 平台使用）
   supported_model_scopes?: string[]
@@ -497,6 +499,7 @@ export interface CreateGroupRequest {
   fallback_group_id?: number | null
   fallback_group_id_on_invalid_request?: number | null
   mcp_xml_inject?: boolean
+  simulate_claude_max_enabled?: boolean
   supported_model_scopes?: string[]
   // 从指定分组复制账号
   copy_accounts_from_group_ids?: number[]
@@ -525,6 +528,7 @@ export interface UpdateGroupRequest {
   fallback_group_id?: number | null
   fallback_group_id_on_invalid_request?: number | null
   mcp_xml_inject?: boolean
+  simulate_claude_max_enabled?: boolean
   supported_model_scopes?: string[]
   copy_accounts_from_group_ids?: number[]
 }
@@ -720,6 +724,12 @@ export interface Account {
   cache_ttl_override_enabled?: boolean | null
   cache_ttl_override_target?: string | null
 
+  // 客户端亲和调度（仅 Anthropic/Antigravity 平台有效）
+  // 启用后新会话会优先调度到客户端之前使用过的账号
+  client_affinity_enabled?: boolean | null
+  affinity_client_count?: number | null
+  affinity_clients?: string[] | null
+
   // API Key 账号配额限制
   quota_limit?: number | null
   quota_used?: number | null
@@ -780,6 +790,11 @@ export interface AccountUsageInfo {
   gemini_pro_minute?: UsageProgress | null
   gemini_flash_minute?: UsageProgress | null
   antigravity_quota?: Record<string, AntigravityModelQuota> | null
+  ai_credits?: Array<{
+    credit_type?: string
+    amount?: number
+    minimum_balance?: number
+  }> | null
   // Antigravity 403 forbidden 状态
   is_forbidden?: boolean
   forbidden_reason?: string