Merge branch 'Wei-Shaw:main' into main

.github/workflows/security-scan.yml

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

backend/.gosec.json

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

backend/internal/handler/admin/account_data.go

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

backend/internal/handler/admin/account_handler.go

@@ -157,7 +157,12 @@ func (h *AccountHandler) List(c *gin.Context) {
 		search = search[:100]
 	}
 
-	accounts, total, err := h.adminService.ListAccounts(c.Request.Context(), page, pageSize, platform, accountType, status, search)
+	var groupID int64
+	if groupIDStr := c.Query("group"); groupIDStr != "" {
+		groupID, _ = strconv.ParseInt(groupIDStr, 10, 64)
+	}
+
+	accounts, total, err := h.adminService.ListAccounts(c.Request.Context(), page, pageSize, platform, accountType, status, search, groupID)
 	if err != nil {
 		response.ErrorFrom(c, err)
 		return
@@ -1413,7 +1418,7 @@ func (h *AccountHandler) BatchRefreshTier(c *gin.Context) {
 	accounts := make([]*service.Account, 0)
 
 	if len(req.AccountIDs) == 0 {
-		allAccounts, _, err := h.adminService.ListAccounts(ctx, 1, 10000, "gemini", "oauth", "", "")
+		allAccounts, _, err := h.adminService.ListAccounts(ctx, 1, 10000, "gemini", "oauth", "", "", 0)
 		if err != nil {
 			response.ErrorFrom(c, err)
 			return

backend/internal/handler/admin/admin_service_stub_test.go

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

backend/internal/repository/account_repo.go

@@ -435,10 +435,10 @@ func (r *accountRepository) Delete(ctx context.Context, id int64) error {
 }
 
 func (r *accountRepository) List(ctx context.Context, params pagination.PaginationParams) ([]service.Account, *pagination.PaginationResult, error) {
-	return r.ListWithFilters(ctx, params, "", "", "", "")
+	return r.ListWithFilters(ctx, params, "", "", "", "", 0)
 }
 
-func (r *accountRepository) ListWithFilters(ctx context.Context, params pagination.PaginationParams, platform, accountType, status, search string) ([]service.Account, *pagination.PaginationResult, error) {
+func (r *accountRepository) ListWithFilters(ctx context.Context, params pagination.PaginationParams, platform, accountType, status, search string, groupID int64) ([]service.Account, *pagination.PaginationResult, error) {
 	q := r.client.Account.Query()
 
 	if platform != "" {
@@ -458,6 +458,9 @@ func (r *accountRepository) ListWithFilters(ctx context.Context, params paginati
 	if search != "" {
 		q = q.Where(dbaccount.NameContainsFold(search))
 	}
+	if groupID > 0 {
+		q = q.Where(dbaccount.HasAccountGroupsWith(dbaccountgroup.GroupIDEQ(groupID)))
+	}
 
 	total, err := q.Count(ctx)
 	if err != nil {

backend/internal/repository/account_repo_integration_test.go

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

backend/internal/server/api_contract_test.go

@@ -936,7 +936,7 @@ func (s *stubAccountRepo) List(ctx context.Context, params pagination.Pagination
 	return nil, nil, errors.New("not implemented")
 }
 
-func (s *stubAccountRepo) ListWithFilters(ctx context.Context, params pagination.PaginationParams, platform, accountType, status, search string) ([]service.Account, *pagination.PaginationResult, error) {
+func (s *stubAccountRepo) ListWithFilters(ctx context.Context, params pagination.PaginationParams, platform, accountType, status, search string, groupID int64) ([]service.Account, *pagination.PaginationResult, error) {
 	return nil, nil, errors.New("not implemented")
 }
 

backend/internal/service/account_service.go

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

backend/internal/service/account_service_delete_test.go

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

backend/internal/service/admin_service.go

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

backend/internal/service/admin_service_search_test.go

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

backend/internal/service/antigravity_gateway_service.go

@@ -4119,6 +4119,15 @@ func (s *AntigravityGatewayService) extractSSEUsage(line string, usage *ClaudeUs
 	if v, ok := u["cache_creation_input_tokens"].(float64); ok && int(v) > 0 {
 		usage.CacheCreationInputTokens = int(v)
 	}
+	// 解析嵌套的 cache_creation 对象中的 5m/1h 明细
+	if cc, ok := u["cache_creation"].(map[string]any); ok {
+		if v, ok := cc["ephemeral_5m_input_tokens"].(float64); ok {
+			usage.CacheCreation5mTokens = int(v)
+		}
+		if v, ok := cc["ephemeral_1h_input_tokens"].(float64); ok {
+			usage.CacheCreation1hTokens = int(v)
+		}
+	}
 }
 
 // extractClaudeUsage 从非流式 Claude 响应提取 usage
@@ -4141,6 +4150,15 @@ func (s *AntigravityGatewayService) extractClaudeUsage(body []byte) *ClaudeUsage
 		if v, ok := u["cache_creation_input_tokens"].(float64); ok {
 			usage.CacheCreationInputTokens = int(v)
 		}
+		// 解析嵌套的 cache_creation 对象中的 5m/1h 明细
+		if cc, ok := u["cache_creation"].(map[string]any); ok {
+			if v, ok := cc["ephemeral_5m_input_tokens"].(float64); ok {
+				usage.CacheCreation5mTokens = int(v)
+			}
+			if v, ok := cc["ephemeral_1h_input_tokens"].(float64); ok {
+				usage.CacheCreation1hTokens = int(v)
+			}
+		}
 	}
 	return usage
 }

backend/internal/service/billing_service.go

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

backend/internal/service/gateway_multiplatform_test.go

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

backend/internal/service/gateway_service.go

@@ -349,6 +349,8 @@ type ClaudeUsage struct {
 	OutputTokens             int `json:"output_tokens"`
 	CacheCreationInputTokens int `json:"cache_creation_input_tokens"`
 	CacheReadInputTokens     int `json:"cache_read_input_tokens"`
+	CacheCreation5mTokens    int // 5分钟缓存创建token（来自嵌套 cache_creation 对象）
+	CacheCreation1hTokens    int // 1小时缓存创建token（来自嵌套 cache_creation 对象）
 }
 
 // ForwardResult 转发结果
@@ -4403,6 +4405,14 @@ func (s *GatewayService) parseSSEUsage(data string, usage *ClaudeUsage) {
 		usage.InputTokens = msgStart.Message.Usage.InputTokens
 		usage.CacheCreationInputTokens = msgStart.Message.Usage.CacheCreationInputTokens
 		usage.CacheReadInputTokens = msgStart.Message.Usage.CacheReadInputTokens
+
+		// 解析嵌套的 cache_creation 对象中的 5m/1h 明细
+		cc5m := gjson.Get(data, "message.usage.cache_creation.ephemeral_5m_input_tokens")
+		cc1h := gjson.Get(data, "message.usage.cache_creation.ephemeral_1h_input_tokens")
+		if cc5m.Exists() || cc1h.Exists() {
+			usage.CacheCreation5mTokens = int(cc5m.Int())
+			usage.CacheCreation1hTokens = int(cc1h.Int())
+		}
 	}
 
 	// 解析message_delta获取tokens（兼容GLM等把所有usage放在delta中的API）
@@ -4431,6 +4441,14 @@ func (s *GatewayService) parseSSEUsage(data string, usage *ClaudeUsage) {
 		if msgDelta.Usage.CacheReadInputTokens > 0 {
 			usage.CacheReadInputTokens = msgDelta.Usage.CacheReadInputTokens
 		}
+
+		// 解析嵌套的 cache_creation 对象中的 5m/1h 明细
+		cc5m := gjson.Get(data, "usage.cache_creation.ephemeral_5m_input_tokens")
+		cc1h := gjson.Get(data, "usage.cache_creation.ephemeral_1h_input_tokens")
+		if cc5m.Exists() || cc1h.Exists() {
+			usage.CacheCreation5mTokens = int(cc5m.Int())
+			usage.CacheCreation1hTokens = int(cc1h.Int())
+		}
 	}
 }
 
@@ -4451,6 +4469,14 @@ func (s *GatewayService) handleNonStreamingResponse(ctx context.Context, resp *h
 		return nil, fmt.Errorf("parse response: %w", err)
 	}
 
+	// 解析嵌套的 cache_creation 对象中的 5m/1h 明细
+	cc5m := gjson.GetBytes(body, "usage.cache_creation.ephemeral_5m_input_tokens")
+	cc1h := gjson.GetBytes(body, "usage.cache_creation.ephemeral_1h_input_tokens")
+	if cc5m.Exists() || cc1h.Exists() {
+		response.Usage.CacheCreation5mTokens = int(cc5m.Int())
+		response.Usage.CacheCreation1hTokens = int(cc1h.Int())
+	}
+
 	// 兼容 Kimi cached_tokens → cache_read_input_tokens
 	if response.Usage.CacheReadInputTokens == 0 {
 		cachedTokens := gjson.GetBytes(body, "usage.cached_tokens").Int()
@@ -4560,10 +4586,12 @@ func (s *GatewayService) RecordUsage(ctx context.Context, input *RecordUsageInpu
 	} else {
 		// Token 计费
 		tokens := UsageTokens{
-			InputTokens:         result.Usage.InputTokens,
+			InputTokens:           result.Usage.InputTokens,
-			OutputTokens:        result.Usage.OutputTokens,
+			OutputTokens:          result.Usage.OutputTokens,
-			CacheCreationTokens: result.Usage.CacheCreationInputTokens,
+			CacheCreationTokens:   result.Usage.CacheCreationInputTokens,
-			CacheReadTokens:     result.Usage.CacheReadInputTokens,
+			CacheReadTokens:       result.Usage.CacheReadInputTokens,
+			CacheCreation5mTokens: result.Usage.CacheCreation5mTokens,
+			CacheCreation1hTokens: result.Usage.CacheCreation1hTokens,
 		}
 		var err error
 		cost, err = s.billingService.CalculateCost(result.Model, tokens, multiplier)
@@ -4597,6 +4625,8 @@ func (s *GatewayService) RecordUsage(ctx context.Context, input *RecordUsageInpu
 		OutputTokens:          result.Usage.OutputTokens,
 		CacheCreationTokens:   result.Usage.CacheCreationInputTokens,
 		CacheReadTokens:       result.Usage.CacheReadInputTokens,
+		CacheCreation5mTokens: result.Usage.CacheCreation5mTokens,
+		CacheCreation1hTokens: result.Usage.CacheCreation1hTokens,
 		InputCost:             cost.InputCost,
 		OutputCost:            cost.OutputCost,
 		CacheCreationCost:     cost.CacheCreationCost,
@@ -4741,10 +4771,12 @@ func (s *GatewayService) RecordUsageWithLongContext(ctx context.Context, input *
 	} else {
 		// Token 计费（使用长上下文计费方法）
 		tokens := UsageTokens{
-			InputTokens:         result.Usage.InputTokens,
+			InputTokens:           result.Usage.InputTokens,
-			OutputTokens:        result.Usage.OutputTokens,
+			OutputTokens:          result.Usage.OutputTokens,
-			CacheCreationTokens: result.Usage.CacheCreationInputTokens,
+			CacheCreationTokens:   result.Usage.CacheCreationInputTokens,
-			CacheReadTokens:     result.Usage.CacheReadInputTokens,
+			CacheReadTokens:       result.Usage.CacheReadInputTokens,
+			CacheCreation5mTokens: result.Usage.CacheCreation5mTokens,
+			CacheCreation1hTokens: result.Usage.CacheCreation1hTokens,
 		}
 		var err error
 		cost, err = s.billingService.CalculateCostWithLongContext(result.Model, tokens, multiplier, input.LongContextThreshold, input.LongContextMultiplier)
@@ -4778,6 +4810,8 @@ func (s *GatewayService) RecordUsageWithLongContext(ctx context.Context, input *
 		OutputTokens:          result.Usage.OutputTokens,
 		CacheCreationTokens:   result.Usage.CacheCreationInputTokens,
 		CacheReadTokens:       result.Usage.CacheReadInputTokens,
+		CacheCreation5mTokens: result.Usage.CacheCreation5mTokens,
+		CacheCreation1hTokens: result.Usage.CacheCreation1hTokens,
 		InputCost:             cost.InputCost,
 		OutputCost:            cost.OutputCost,
 		CacheCreationCost:     cost.CacheCreationCost,

backend/internal/service/gemini_multiplatform_test.go

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

backend/internal/service/ops_concurrency.go

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

backend/internal/service/pricing_service.go

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

backend/internal/service/ratelimit_service.go

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

backend/internal/service/ratelimit_service_anthropic_test.go

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

backend/internal/service/usage_log.go

@@ -26,8 +26,8 @@ type UsageLog struct {
 	CacheCreationTokens int
 	CacheReadTokens     int
 
-	CacheCreation5mTokens int
+	CacheCreation5mTokens int `gorm:"column:cache_creation_5m_tokens"`
-	CacheCreation1hTokens int
+	CacheCreation1hTokens int `gorm:"column:cache_creation_1h_tokens"`
 
 	InputCost         float64
 	OutputCost        float64

backend/migrations/054_drop_legacy_cache_columns.sql

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

deploy/docker-compose.yml

@@ -164,6 +164,7 @@ services:
       - POSTGRES_USER=${POSTGRES_USER:-sub2api}
       - POSTGRES_PASSWORD=${POSTGRES_PASSWORD:?POSTGRES_PASSWORD is required}
       - POSTGRES_DB=${POSTGRES_DB:-sub2api}
+      - PGDATA=/var/lib/postgresql/data
       - TZ=${TZ:-Asia/Shanghai}
     networks:
       - sub2api-network

frontend/src/api/admin/accounts.ts

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

frontend/src/components/account/AccountGroupsCell.vue

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

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

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

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

@@ -70,6 +70,7 @@
                 <div v-if="row.cache_creation_tokens > 0" class="inline-flex items-center gap-1">
                   <svg class="h-3.5 w-3.5 text-amber-500" fill="none" stroke="currentColor" viewBox="0 0 24 24"><path stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M11 5H6a2 2 0 00-2 2v11a2 2 0 002 2h11a2 2 0 002-2v-5m-1.414-9.414a2 2 0 112.828 2.828L11.828 15H9v-2.828l8.586-8.586z" /></svg>
                   <span class="font-medium text-amber-600 dark:text-amber-400">{{ formatCacheTokens(row.cache_creation_tokens) }}</span>
+                  <span v-if="row.cache_creation_1h_tokens > 0" class="inline-flex items-center rounded px-1 py-px text-[10px] font-medium leading-tight bg-orange-100 text-orange-600 ring-1 ring-inset ring-orange-200 dark:bg-orange-500/20 dark:text-orange-400 dark:ring-orange-500/30">1h</span>
                 </div>
               </div>
             </div>
@@ -157,9 +158,29 @@
               <span class="text-gray-400">{{ t('admin.usage.outputTokens') }}</span>
               <span class="font-medium text-white">{{ tokenTooltipData.output_tokens.toLocaleString() }}</span>
             </div>
-            <div v-if="tokenTooltipData && tokenTooltipData.cache_creation_tokens > 0" class="flex items-center justify-between gap-4">
+            <div v-if="tokenTooltipData && tokenTooltipData.cache_creation_tokens > 0">
-              <span class="text-gray-400">{{ t('admin.usage.cacheCreationTokens') }}</span>
+              <!-- 有 5m/1h 明细时，展开显示 -->
-              <span class="font-medium text-white">{{ tokenTooltipData.cache_creation_tokens.toLocaleString() }}</span>
+              <template v-if="tokenTooltipData.cache_creation_5m_tokens > 0 || tokenTooltipData.cache_creation_1h_tokens > 0">
+                <div v-if="tokenTooltipData.cache_creation_5m_tokens > 0" class="flex items-center justify-between gap-4">
+                  <span class="text-gray-400 flex items-center gap-1.5">
+                    {{ t('admin.usage.cacheCreation5mTokens') }}
+                    <span class="inline-flex items-center rounded px-1 py-px text-[10px] font-medium leading-tight bg-amber-500/20 text-amber-400 ring-1 ring-inset ring-amber-500/30">5m</span>
+                  </span>
+                  <span class="font-medium text-white">{{ tokenTooltipData.cache_creation_5m_tokens.toLocaleString() }}</span>
+                </div>
+                <div v-if="tokenTooltipData.cache_creation_1h_tokens > 0" class="flex items-center justify-between gap-4">
+                  <span class="text-gray-400 flex items-center gap-1.5">
+                    {{ t('admin.usage.cacheCreation1hTokens') }}
+                    <span class="inline-flex items-center rounded px-1 py-px text-[10px] font-medium leading-tight bg-orange-500/20 text-orange-400 ring-1 ring-inset ring-orange-500/30">1h</span>
+                  </span>
+                  <span class="font-medium text-white">{{ tokenTooltipData.cache_creation_1h_tokens.toLocaleString() }}</span>
+                </div>
+              </template>
+              <!-- 无明细时，只显示聚合值 -->
+              <div v-else class="flex items-center justify-between gap-4">
+                <span class="text-gray-400">{{ t('admin.usage.cacheCreationTokens') }}</span>
+                <span class="font-medium text-white">{{ tokenTooltipData.cache_creation_tokens.toLocaleString() }}</span>
+              </div>
             </div>
             <div v-if="tokenTooltipData && tokenTooltipData.cache_read_tokens > 0" class="flex items-center justify-between gap-4">
               <span class="text-gray-400">{{ t('admin.usage.cacheReadTokens') }}</span>

frontend/src/components/common/StatCard.vue

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

frontend/src/components/layout/AppSidebar.vue

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

frontend/src/i18n/locales/en.ts

@@ -1335,6 +1335,7 @@ export default {
       allPlatforms: 'All Platforms',
       allTypes: 'All Types',
       allStatus: 'All Status',
+      allGroups: 'All Groups',
       oauthType: 'OAuth',
       setupToken: 'Setup Token',
       apiKey: 'API Key',
@@ -1344,7 +1345,7 @@ export default {
       schedulableEnabled: 'Scheduling enabled',
       schedulableDisabled: 'Scheduling disabled',
       failedToToggleSchedulable: 'Failed to toggle scheduling status',
-      allGroups: '{count} groups total',
+      groupCountTotal: '{count} groups total',
       platforms: {
         anthropic: 'Anthropic',
         claude: 'Claude',
@@ -2359,6 +2360,8 @@ export default {
       inputTokens: 'Input Tokens',
       outputTokens: 'Output Tokens',
       cacheCreationTokens: 'Cache Creation Tokens',
+      cacheCreation5mTokens: 'Cache Write',
+      cacheCreation1hTokens: 'Cache Write',
       cacheReadTokens: 'Cache Read Tokens',
       failedToLoad: 'Failed to load usage records',
       billingType: 'Billing Type',

frontend/src/i18n/locales/zh.ts

@@ -1426,6 +1426,7 @@ export default {
       allPlatforms: '全部平台',
       allTypes: '全部类型',
       allStatus: '全部状态',
+      allGroups: '全部分组',
       oauthType: 'OAuth',
       // Schedulable toggle
       schedulable: '参与调度',
@@ -1433,7 +1434,7 @@ export default {
       schedulableEnabled: '调度已开启',
       schedulableDisabled: '调度已关闭',
       failedToToggleSchedulable: '切换调度状态失败',
-      allGroups: '共 {count} 个分组',
+      groupCountTotal: '共 {count} 个分组',
       columns: {
         name: '名称',
         platformType: '平台/类型',
@@ -2526,6 +2527,8 @@ export default {
       inputTokens: '输入 Token',
       outputTokens: '输出 Token',
       cacheCreationTokens: '缓存创建 Token',
+      cacheCreation5mTokens: '缓存创建',
+      cacheCreation1hTokens: '缓存创建',
       cacheReadTokens: '缓存读取 Token',
       failedToLoad: '加载使用记录失败',
       billingType: '计费类型',

frontend/src/style.css

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

frontend/src/views/admin/AccountsView.vue

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

frontend/src/views/user/UsageView.vue

@@ -233,6 +233,7 @@
                     <span class="font-medium text-amber-600 dark:text-amber-400">{{
                       formatCacheTokens(row.cache_creation_tokens)
                     }}</span>
+                    <span v-if="row.cache_creation_1h_tokens > 0" class="inline-flex items-center rounded px-1 py-px text-[10px] font-medium leading-tight bg-orange-100 text-orange-600 ring-1 ring-inset ring-orange-200 dark:bg-orange-500/20 dark:text-orange-400 dark:ring-orange-500/30">1h</span>
                   </div>
                 </div>
               </div>
@@ -350,9 +351,29 @@
               <span class="text-gray-400">{{ t('admin.usage.outputTokens') }}</span>
               <span class="font-medium text-white">{{ tokenTooltipData.output_tokens.toLocaleString() }}</span>
             </div>
-            <div v-if="tokenTooltipData && tokenTooltipData.cache_creation_tokens > 0" class="flex items-center justify-between gap-4">
+            <div v-if="tokenTooltipData && tokenTooltipData.cache_creation_tokens > 0">
-              <span class="text-gray-400">{{ t('admin.usage.cacheCreationTokens') }}</span>
+              <!-- 有 5m/1h 明细时，展开显示 -->
-              <span class="font-medium text-white">{{ tokenTooltipData.cache_creation_tokens.toLocaleString() }}</span>
+              <template v-if="tokenTooltipData.cache_creation_5m_tokens > 0 || tokenTooltipData.cache_creation_1h_tokens > 0">
+                <div v-if="tokenTooltipData.cache_creation_5m_tokens > 0" class="flex items-center justify-between gap-4">
+                  <span class="text-gray-400 flex items-center gap-1.5">
+                    {{ t('admin.usage.cacheCreation5mTokens') }}
+                    <span class="inline-flex items-center rounded px-1 py-px text-[10px] font-medium leading-tight bg-amber-500/20 text-amber-400 ring-1 ring-inset ring-amber-500/30">5m</span>
+                  </span>
+                  <span class="font-medium text-white">{{ tokenTooltipData.cache_creation_5m_tokens.toLocaleString() }}</span>
+                </div>
+                <div v-if="tokenTooltipData.cache_creation_1h_tokens > 0" class="flex items-center justify-between gap-4">
+                  <span class="text-gray-400 flex items-center gap-1.5">
+                    {{ t('admin.usage.cacheCreation1hTokens') }}
+                    <span class="inline-flex items-center rounded px-1 py-px text-[10px] font-medium leading-tight bg-orange-500/20 text-orange-400 ring-1 ring-inset ring-orange-500/30">1h</span>
+                  </span>
+                  <span class="font-medium text-white">{{ tokenTooltipData.cache_creation_1h_tokens.toLocaleString() }}</span>
+                </div>
+              </template>
+              <!-- 无明细时，只显示聚合值 -->
+              <div v-else class="flex items-center justify-between gap-4">
+                <span class="text-gray-400">{{ t('admin.usage.cacheCreationTokens') }}</span>
+                <span class="font-medium text-white">{{ tokenTooltipData.cache_creation_tokens.toLocaleString() }}</span>
+              </div>
             </div>
             <div v-if="tokenTooltipData && tokenTooltipData.cache_read_tokens > 0" class="flex items-center justify-between gap-4">
               <span class="text-gray-400">{{ t('admin.usage.cacheReadTokens') }}</span>