feat: add quota limit for API key accounts

- Add configurable spending limit (quota_limit) for apikey-type accounts
- Atomic quota accumulation via PostgreSQL JSONB operations on TotalCost
- Scheduler filters out over-quota accounts with outbox-triggered snapshot refresh
- Display quota usage ($used / $limit) in account capacity column
- Add "Reset Quota" action in account menu to reset usage to zero
- Editing account settings preserves quota_used (no accidental reset)
- Covers all 3 billing paths: Anthropic, Gemini, OpenAI RecordUsage

chore: bump version to 0.1.90.4

backend/internal/handler/admin/account_handler.go

@@ -1334,6 +1334,29 @@ func (h *AccountHandler) ClearRateLimit(c *gin.Context) {
 	response.Success(c, h.buildAccountResponseWithRuntime(c.Request.Context(), account))
 }
 
+// ResetQuota handles resetting account quota usage
+// POST /api/v1/admin/accounts/:id/reset-quota
+func (h *AccountHandler) ResetQuota(c *gin.Context) {
+	accountID, err := strconv.ParseInt(c.Param("id"), 10, 64)
+	if err != nil {
+		response.BadRequest(c, "Invalid account ID")
+		return
+	}
+
+	if err := h.adminService.ResetAccountQuota(c.Request.Context(), accountID); err != nil {
+		response.InternalError(c, "Failed to reset account quota: "+err.Error())
+		return
+	}
+
+	account, err := h.adminService.GetAccount(c.Request.Context(), accountID)
+	if err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
+
+	response.Success(c, h.buildAccountResponseWithRuntime(c.Request.Context(), account))
+}
+
 // GetTempUnschedulable handles getting temporary unschedulable status
 // GET /api/v1/admin/accounts/:id/temp-unschedulable
 func (h *AccountHandler) GetTempUnschedulable(c *gin.Context) {

backend/internal/handler/admin/admin_service_stub_test.go

@@ -425,5 +425,9 @@ func (s *stubAdminService) AdminUpdateAPIKeyGroupID(ctx context.Context, keyID i
 	return nil, service.ErrAPIKeyNotFound
 }
 
+func (s *stubAdminService) ResetAccountQuota(ctx context.Context, id int64) error {
+	return nil
+}
+
 // Ensure stub implements interface.
 var _ service.AdminService = (*stubAdminService)(nil)

backend/internal/handler/dto/mappers.go

@@ -249,6 +249,17 @@ func AccountFromServiceShallow(a *service.Account) *Account {
 		}
 	}
 
+	// 提取 API Key 账号配额限制（仅 apikey 类型有效）
+	if a.Type == service.AccountTypeAPIKey {
+		if limit := a.GetQuotaLimit(); limit > 0 {
+			out.QuotaLimit = &limit
+		}
+		used := a.GetQuotaUsed()
+		if out.QuotaLimit != nil {
+			out.QuotaUsed = &used
+		}
+	}
+
 	return out
 }
 

backend/internal/handler/dto/types.go

@@ -187,6 +187,10 @@ type Account struct {
 	CacheTTLOverrideEnabled *bool   `json:"cache_ttl_override_enabled,omitempty"`
 	CacheTTLOverrideTarget  *string `json:"cache_ttl_override_target,omitempty"`
 
+	// API Key 账号配额限制
+	QuotaLimit *float64 `json:"quota_limit,omitempty"`
+	QuotaUsed  *float64 `json:"quota_used,omitempty"`
+
 	Proxy         *Proxy         `json:"proxy,omitempty"`
 	AccountGroups []AccountGroup `json:"account_groups,omitempty"`
 

backend/internal/handler/sora_client_handler_test.go

@@ -2132,6 +2132,14 @@ func (r *stubAccountRepoForHandler) BulkUpdate(context.Context, []int64, service
 	return 0, nil
 }
 
+func (r *stubAccountRepoForHandler) IncrementQuotaUsed(context.Context, int64, float64) error {
+	return nil
+}
+
+func (r *stubAccountRepoForHandler) ResetQuotaUsed(context.Context, int64) error {
+	return nil
+}
+
 // ==================== Stub: SoraClient (用于 SoraGatewayService) ====================
 
 var _ service.SoraClient = (*stubSoraClientForHandler)(nil)

backend/internal/handler/sora_gateway_handler_test.go

@@ -216,6 +216,14 @@ func (r *stubAccountRepo) BulkUpdate(ctx context.Context, ids []int64, updates s
 	return 0, nil
 }
 
+func (r *stubAccountRepo) IncrementQuotaUsed(ctx context.Context, id int64, amount float64) error {
+	return nil
+}
+
+func (r *stubAccountRepo) ResetQuotaUsed(ctx context.Context, id int64) error {
+	return nil
+}
+
 func (r *stubAccountRepo) listSchedulable() []service.Account {
 	var result []service.Account
 	for _, acc := range r.accounts {

backend/internal/repository/account_repo.go

@@ -1639,3 +1639,60 @@ func (r *accountRepository) FindByExtraField(ctx context.Context, key string, va
 
 	return r.accountsToService(ctx, accounts)
 }
+
+// IncrementQuotaUsed 原子递增账号的 extra.quota_used 字段
+func (r *accountRepository) IncrementQuotaUsed(ctx context.Context, id int64, amount float64) error {
+	rows, err := r.sql.QueryContext(ctx,
+		`UPDATE accounts SET extra = jsonb_set(
+			COALESCE(extra, '{}'::jsonb),
+			'{quota_used}',
+			to_jsonb(COALESCE((extra->>'quota_used')::numeric, 0) + $1)
+		), updated_at = NOW()
+		WHERE id = $2 AND deleted_at IS NULL
+		RETURNING
+			COALESCE((extra->>'quota_used')::numeric, 0),
+			COALESCE((extra->>'quota_limit')::numeric, 0)`,
+		amount, id)
+	if err != nil {
+		return err
+	}
+	defer func() { _ = rows.Close() }()
+
+	var newUsed, limit float64
+	if rows.Next() {
+		if err := rows.Scan(&newUsed, &limit); err != nil {
+			return err
+		}
+	}
+	if err := rows.Err(); err != nil {
+		return err
+	}
+
+	// 配额刚超限时触发调度快照刷新，使账号及时从调度候选中移除
+	if limit > 0 && newUsed >= limit && (newUsed-amount) < limit {
+		if err := enqueueSchedulerOutbox(ctx, r.sql, service.SchedulerOutboxEventAccountChanged, &id, nil, nil); err != nil {
+			logger.LegacyPrintf("repository.account", "[SchedulerOutbox] enqueue quota exceeded failed: account=%d err=%v", id, err)
+		}
+	}
+	return nil
+}
+
+// ResetQuotaUsed 重置账号的 extra.quota_used 为 0
+func (r *accountRepository) ResetQuotaUsed(ctx context.Context, id int64) error {
+	_, err := r.sql.ExecContext(ctx,
+		`UPDATE accounts SET extra = jsonb_set(
+			COALESCE(extra, '{}'::jsonb),
+			'{quota_used}',
+			'0'::jsonb
+		), updated_at = NOW()
+		WHERE id = $1 AND deleted_at IS NULL`,
+		id)
+	if err != nil {
+		return err
+	}
+	// 重置配额后触发调度快照刷新，使账号重新参与调度
+	if err := enqueueSchedulerOutbox(ctx, r.sql, service.SchedulerOutboxEventAccountChanged, &id, nil, nil); err != nil {
+		logger.LegacyPrintf("repository.account", "[SchedulerOutbox] enqueue quota reset failed: account=%d err=%v", id, err)
+	}
+	return nil
+}

backend/internal/server/routes/admin.go

@@ -249,6 +249,7 @@ func registerAccountRoutes(admin *gin.RouterGroup, h *handler.Handlers) {
 		accounts.GET("/:id/today-stats", h.Admin.Account.GetTodayStats)
 		accounts.POST("/today-stats/batch", h.Admin.Account.GetBatchTodayStats)
 		accounts.POST("/:id/clear-rate-limit", h.Admin.Account.ClearRateLimit)
+		accounts.POST("/:id/reset-quota", h.Admin.Account.ResetQuota)
 		accounts.GET("/:id/temp-unschedulable", h.Admin.Account.GetTempUnschedulable)
 		accounts.DELETE("/:id/temp-unschedulable", h.Admin.Account.ClearTempUnschedulable)
 		accounts.POST("/:id/schedulable", h.Admin.Account.SetSchedulable)

backend/internal/service/account.go

@@ -1104,6 +1104,38 @@ func (a *Account) GetCacheTTLOverrideTarget() string {
 	return "5m"
 }
 
+// GetQuotaLimit 获取 API Key 账号的配额限制（美元）
+// 返回 0 表示未启用
+func (a *Account) GetQuotaLimit() float64 {
+	if a.Extra == nil {
+		return 0
+	}
+	if v, ok := a.Extra["quota_limit"]; ok {
+		return parseExtraFloat64(v)
+	}
+	return 0
+}
+
+// GetQuotaUsed 获取 API Key 账号的已用配额（美元）
+func (a *Account) GetQuotaUsed() float64 {
+	if a.Extra == nil {
+		return 0
+	}
+	if v, ok := a.Extra["quota_used"]; ok {
+		return parseExtraFloat64(v)
+	}
+	return 0
+}
+
+// IsQuotaExceeded 检查 API Key 账号配额是否已超限
+func (a *Account) IsQuotaExceeded() bool {
+	limit := a.GetQuotaLimit()
+	if limit <= 0 {
+		return false
+	}
+	return a.GetQuotaUsed() >= limit
+}
+
 // GetWindowCostLimit 获取 5h 窗口费用阈值（美元）
 // 返回 0 表示未启用
 func (a *Account) GetWindowCostLimit() float64 {

backend/internal/service/account_service.go

@@ -68,6 +68,10 @@ type AccountRepository interface {
 	UpdateSessionWindow(ctx context.Context, id int64, start, end *time.Time, status string) error
 	UpdateExtra(ctx context.Context, id int64, updates map[string]any) error
 	BulkUpdate(ctx context.Context, ids []int64, updates AccountBulkUpdate) (int64, error)
+	// IncrementQuotaUsed 原子递增 API Key 账号的配额用量
+	IncrementQuotaUsed(ctx context.Context, id int64, amount float64) error
+	// ResetQuotaUsed 重置 API Key 账号的配额用量为 0
+	ResetQuotaUsed(ctx context.Context, id int64) error
 }
 
 // AccountBulkUpdate describes the fields that can be updated in a bulk operation.

backend/internal/service/account_service_delete_test.go

@@ -199,6 +199,14 @@ func (s *accountRepoStub) BulkUpdate(ctx context.Context, ids []int64, updates A
 	panic("unexpected BulkUpdate call")
 }
 
+func (s *accountRepoStub) IncrementQuotaUsed(ctx context.Context, id int64, amount float64) error {
+	return nil
+}
+
+func (s *accountRepoStub) ResetQuotaUsed(ctx context.Context, id int64) error {
+	return nil
+}
+
 // TestAccountService_Delete_NotFound 测试删除不存在的账号时返回正确的错误。
 // 预期行为：
 //   - ExistsByID 返回 false（账号不存在）

backend/internal/service/admin_service.go

@@ -84,6 +84,7 @@ type AdminService interface {
 	DeleteRedeemCode(ctx context.Context, id int64) error
 	BatchDeleteRedeemCodes(ctx context.Context, ids []int64) (int64, error)
 	ExpireRedeemCode(ctx context.Context, id int64) (*RedeemCode, error)
+	ResetAccountQuota(ctx context.Context, id int64) error
 }
 
 // CreateUserInput represents input for creating a new user via admin operations.
@@ -1477,6 +1478,10 @@ func (s *adminServiceImpl) UpdateAccount(ctx context.Context, id int64, input *U
 		account.Credentials = input.Credentials
 	}
 	if len(input.Extra) > 0 {
+		// 保留 quota_used，防止编辑账号时意外重置配额用量
+		if oldQuotaUsed, ok := account.Extra["quota_used"]; ok {
+			input.Extra["quota_used"] = oldQuotaUsed
+		}
 		account.Extra = input.Extra
 	}
 	if input.ProxyID != nil {
@@ -2458,3 +2463,7 @@ func (e *MixedChannelError) Error() string {
 	return fmt.Sprintf("mixed_channel_warning: Group '%s' contains both %s and %s accounts. Using mixed channels in the same context may cause thinking block signature validation issues, which will fallback to non-thinking mode for historical messages.",
 		e.GroupName, e.CurrentPlatform, e.OtherPlatform)
 }
+
+func (s *adminServiceImpl) ResetAccountQuota(ctx context.Context, id int64) error {
+	return s.accountRepo.ResetQuotaUsed(ctx, id)
+}

backend/internal/service/gateway_multiplatform_test.go

@@ -187,6 +187,14 @@ func (m *mockAccountRepoForPlatform) BulkUpdate(ctx context.Context, ids []int64
 	return 0, nil
 }
 
+func (m *mockAccountRepoForPlatform) IncrementQuotaUsed(ctx context.Context, id int64, amount float64) error {
+	return nil
+}
+
+func (m *mockAccountRepoForPlatform) ResetQuotaUsed(ctx context.Context, id int64) error {
+	return nil
+}
+
 // Verify interface implementation
 var _ AccountRepository = (*mockAccountRepoForPlatform)(nil)
 

backend/internal/service/gateway_service.go

@@ -1234,6 +1234,10 @@ func (s *GatewayService) SelectAccountWithLoadAwareness(ctx context.Context, gro
 				modelScopeSkippedIDs = append(modelScopeSkippedIDs, account.ID)
 				continue
 			}
+			// 配额检查
+			if !s.isAccountSchedulableForQuota(account) {
+				continue
+			}
 			// 窗口费用检查（非粘性会话路径）
 			if !s.isAccountSchedulableForWindowCost(ctx, account, false) {
 				filteredWindowCost++
@@ -1266,6 +1270,7 @@ func (s *GatewayService) SelectAccountWithLoadAwareness(ctx context.Context, gro
 							s.isAccountAllowedForPlatform(stickyAccount, platform, useMixed) &&
 							(requestedModel == "" || s.isModelSupportedByAccountWithContext(ctx, stickyAccount, requestedModel)) &&
 							s.isAccountSchedulableForModelSelection(ctx, stickyAccount, requestedModel) &&
+							s.isAccountSchedulableForQuota(stickyAccount) &&
 							s.isAccountSchedulableForWindowCost(ctx, stickyAccount, true) &&
 
 							s.isAccountSchedulableForRPM(ctx, stickyAccount, true) { // 粘性会话窗口费用+RPM 检查
@@ -1422,6 +1427,7 @@ func (s *GatewayService) SelectAccountWithLoadAwareness(ctx context.Context, gro
 					s.isAccountAllowedForPlatform(account, platform, useMixed) &&
 					(requestedModel == "" || s.isModelSupportedByAccountWithContext(ctx, account, requestedModel)) &&
 					s.isAccountSchedulableForModelSelection(ctx, account, requestedModel) &&
+					s.isAccountSchedulableForQuota(account) &&
 					s.isAccountSchedulableForWindowCost(ctx, account, true) &&
 
 					s.isAccountSchedulableForRPM(ctx, account, true) { // 粘性会话窗口费用+RPM 检查
@@ -1486,6 +1492,10 @@ func (s *GatewayService) SelectAccountWithLoadAwareness(ctx context.Context, gro
 		if !s.isAccountSchedulableForModelSelection(ctx, acc, requestedModel) {
 			continue
 		}
+		// 配额检查
+		if !s.isAccountSchedulableForQuota(acc) {
+			continue
+		}
 		// 窗口费用检查（非粘性会话路径）
 		if !s.isAccountSchedulableForWindowCost(ctx, acc, false) {
 			continue
@@ -2119,6 +2129,15 @@ func (s *GatewayService) withWindowCostPrefetch(ctx context.Context, accounts []
 	return context.WithValue(ctx, windowCostPrefetchContextKey, costs)
 }
 
+// isAccountSchedulableForQuota 检查 API Key 账号是否在配额限制内
+// 仅适用于配置了 quota_limit 的 apikey 类型账号
+func (s *GatewayService) isAccountSchedulableForQuota(account *Account) bool {
+	if account.Type != AccountTypeAPIKey {
+		return true
+	}
+	return !account.IsQuotaExceeded()
+}
+
 // isAccountSchedulableForWindowCost 检查账号是否可根据窗口费用进行调度
 // 仅适用于 Anthropic OAuth/SetupToken 账号
 // 返回 true 表示可调度，false 表示不可调度
@@ -2596,7 +2615,7 @@ func (s *GatewayService) selectAccountForModelWithPlatform(ctx context.Context,
 						if clearSticky {
 							_ = s.cache.DeleteSessionAccountID(ctx, derefGroupID(groupID), sessionHash)
 						}
-						if !clearSticky && s.isAccountInGroup(account, groupID) && account.Platform == platform && (requestedModel == "" || s.isModelSupportedByAccountWithContext(ctx, account, requestedModel)) && s.isAccountSchedulableForModelSelection(ctx, account, requestedModel) && s.isAccountSchedulableForWindowCost(ctx, account, true) && s.isAccountSchedulableForRPM(ctx, account, true) {
+						if !clearSticky && s.isAccountInGroup(account, groupID) && account.Platform == platform && (requestedModel == "" || s.isModelSupportedByAccountWithContext(ctx, account, requestedModel)) && s.isAccountSchedulableForModelSelection(ctx, account, requestedModel) && s.isAccountSchedulableForQuota(account) && s.isAccountSchedulableForWindowCost(ctx, account, true) && s.isAccountSchedulableForRPM(ctx, account, true) {
 							if s.debugModelRoutingEnabled() {
 								logger.LegacyPrintf("service.gateway", "[ModelRoutingDebug] legacy routed sticky hit: group_id=%v model=%s session=%s account=%d", derefGroupID(groupID), requestedModel, shortSessionHash(sessionHash), accountID)
 							}
@@ -2650,6 +2669,9 @@ func (s *GatewayService) selectAccountForModelWithPlatform(ctx context.Context,
 			if !s.isAccountSchedulableForModelSelection(ctx, acc, requestedModel) {
 				continue
 			}
+			if !s.isAccountSchedulableForQuota(acc) {
+				continue
+			}
 			if !s.isAccountSchedulableForWindowCost(ctx, acc, false) {
 				continue
 			}
@@ -2706,7 +2728,7 @@ func (s *GatewayService) selectAccountForModelWithPlatform(ctx context.Context,
 					if clearSticky {
 						_ = s.cache.DeleteSessionAccountID(ctx, derefGroupID(groupID), sessionHash)
 					}
-					if !clearSticky && s.isAccountInGroup(account, groupID) && account.Platform == platform && (requestedModel == "" || s.isModelSupportedByAccountWithContext(ctx, account, requestedModel)) && s.isAccountSchedulableForModelSelection(ctx, account, requestedModel) && s.isAccountSchedulableForWindowCost(ctx, account, true) && s.isAccountSchedulableForRPM(ctx, account, true) {
+					if !clearSticky && s.isAccountInGroup(account, groupID) && account.Platform == platform && (requestedModel == "" || s.isModelSupportedByAccountWithContext(ctx, account, requestedModel)) && s.isAccountSchedulableForModelSelection(ctx, account, requestedModel) && s.isAccountSchedulableForQuota(account) && s.isAccountSchedulableForWindowCost(ctx, account, true) && s.isAccountSchedulableForRPM(ctx, account, true) {
 						return account, nil
 					}
 				}
@@ -2749,6 +2771,9 @@ func (s *GatewayService) selectAccountForModelWithPlatform(ctx context.Context,
 		if !s.isAccountSchedulableForModelSelection(ctx, acc, requestedModel) {
 			continue
 		}
+		if !s.isAccountSchedulableForQuota(acc) {
+			continue
+		}
 		if !s.isAccountSchedulableForWindowCost(ctx, acc, false) {
 			continue
 		}
@@ -2824,7 +2849,7 @@ func (s *GatewayService) selectAccountWithMixedScheduling(ctx context.Context, g
 						if clearSticky {
 							_ = s.cache.DeleteSessionAccountID(ctx, derefGroupID(groupID), sessionHash)
 						}
-						if !clearSticky && s.isAccountInGroup(account, groupID) && (requestedModel == "" || s.isModelSupportedByAccountWithContext(ctx, account, requestedModel)) && s.isAccountSchedulableForModelSelection(ctx, account, requestedModel) && s.isAccountSchedulableForWindowCost(ctx, account, true) && s.isAccountSchedulableForRPM(ctx, account, true) {
+						if !clearSticky && s.isAccountInGroup(account, groupID) && (requestedModel == "" || s.isModelSupportedByAccountWithContext(ctx, account, requestedModel)) && s.isAccountSchedulableForModelSelection(ctx, account, requestedModel) && s.isAccountSchedulableForQuota(account) && s.isAccountSchedulableForWindowCost(ctx, account, true) && s.isAccountSchedulableForRPM(ctx, account, true) {
 							if account.Platform == nativePlatform || (account.Platform == PlatformAntigravity && account.IsMixedSchedulingEnabled()) {
 								if s.debugModelRoutingEnabled() {
 									logger.LegacyPrintf("service.gateway", "[ModelRoutingDebug] legacy mixed routed sticky hit: group_id=%v model=%s session=%s account=%d", derefGroupID(groupID), requestedModel, shortSessionHash(sessionHash), accountID)
@@ -2880,6 +2905,9 @@ func (s *GatewayService) selectAccountWithMixedScheduling(ctx context.Context, g
 			if !s.isAccountSchedulableForModelSelection(ctx, acc, requestedModel) {
 				continue
 			}
+			if !s.isAccountSchedulableForQuota(acc) {
+				continue
+			}
 			if !s.isAccountSchedulableForWindowCost(ctx, acc, false) {
 				continue
 			}
@@ -2936,7 +2964,7 @@ func (s *GatewayService) selectAccountWithMixedScheduling(ctx context.Context, g
 					if clearSticky {
 						_ = s.cache.DeleteSessionAccountID(ctx, derefGroupID(groupID), sessionHash)
 					}
-					if !clearSticky && s.isAccountInGroup(account, groupID) && (requestedModel == "" || s.isModelSupportedByAccountWithContext(ctx, account, requestedModel)) && s.isAccountSchedulableForModelSelection(ctx, account, requestedModel) && s.isAccountSchedulableForWindowCost(ctx, account, true) && s.isAccountSchedulableForRPM(ctx, account, true) {
+					if !clearSticky && s.isAccountInGroup(account, groupID) && (requestedModel == "" || s.isModelSupportedByAccountWithContext(ctx, account, requestedModel)) && s.isAccountSchedulableForModelSelection(ctx, account, requestedModel) && s.isAccountSchedulableForQuota(account) && s.isAccountSchedulableForWindowCost(ctx, account, true) && s.isAccountSchedulableForRPM(ctx, account, true) {
 						if account.Platform == nativePlatform || (account.Platform == PlatformAntigravity && account.IsMixedSchedulingEnabled()) {
 							return account, nil
 						}
@@ -2981,6 +3009,9 @@ func (s *GatewayService) selectAccountWithMixedScheduling(ctx context.Context, g
 		if !s.isAccountSchedulableForModelSelection(ctx, acc, requestedModel) {
 			continue
 		}
+		if !s.isAccountSchedulableForQuota(acc) {
+			continue
+		}
 		if !s.isAccountSchedulableForWindowCost(ctx, acc, false) {
 			continue
 		}
@@ -6601,6 +6632,13 @@ func (s *GatewayService) RecordUsage(ctx context.Context, input *RecordUsageInpu
 		s.billingCacheService.QueueUpdateAPIKeyRateLimitUsage(apiKey.ID, cost.ActualCost)
 	}
 
+	// 更新 API Key 账号配额用量
+	if shouldBill && cost.TotalCost > 0 && account.Type == AccountTypeAPIKey && account.GetQuotaLimit() > 0 {
+		if err := s.accountRepo.IncrementQuotaUsed(ctx, account.ID, cost.TotalCost); err != nil {
+			slog.Error("increment account quota used failed", "account_id", account.ID, "cost", cost.TotalCost, "error", err)
+		}
+	}
+
 	// Schedule batch update for account last_used_at
 	s.deferredService.ScheduleLastUsedUpdate(account.ID)
 
@@ -6798,6 +6836,13 @@ func (s *GatewayService) RecordUsageWithLongContext(ctx context.Context, input *
 		s.billingCacheService.QueueUpdateAPIKeyRateLimitUsage(apiKey.ID, cost.ActualCost)
 	}
 
+	// 更新 API Key 账号配额用量
+	if shouldBill && cost.TotalCost > 0 && account.Type == AccountTypeAPIKey && account.GetQuotaLimit() > 0 {
+		if err := s.accountRepo.IncrementQuotaUsed(ctx, account.ID, cost.TotalCost); err != nil {
+			slog.Error("increment account quota used failed", "account_id", account.ID, "cost", cost.TotalCost, "error", err)
+		}
+	}
+
 	// Schedule batch update for account last_used_at
 	s.deferredService.ScheduleLastUsedUpdate(account.ID)
 

backend/internal/service/gemini_multiplatform_test.go

@@ -176,6 +176,14 @@ func (m *mockAccountRepoForGemini) BulkUpdate(ctx context.Context, ids []int64,
 	return 0, nil
 }
 
+func (m *mockAccountRepoForGemini) IncrementQuotaUsed(ctx context.Context, id int64, amount float64) error {
+	return nil
+}
+
+func (m *mockAccountRepoForGemini) ResetQuotaUsed(ctx context.Context, id int64) error {
+	return nil
+}
+
 // Verify interface implementation
 var _ AccountRepository = (*mockAccountRepoForGemini)(nil)
 

backend/internal/service/openai_gateway_service.go

@@ -3500,6 +3500,13 @@ func (s *OpenAIGatewayService) RecordUsage(ctx context.Context, input *OpenAIRec
 		s.billingCacheService.QueueUpdateAPIKeyRateLimitUsage(apiKey.ID, cost.ActualCost)
 	}
 
+	// 更新 API Key 账号配额用量
+	if shouldBill && cost.TotalCost > 0 && account.Type == AccountTypeAPIKey && account.GetQuotaLimit() > 0 {
+		if err := s.accountRepo.IncrementQuotaUsed(ctx, account.ID, cost.TotalCost); err != nil {
+			logger.LegacyPrintf("service.openai_gateway", "increment account quota used failed: account_id=%d cost=%f error=%v", account.ID, cost.TotalCost, err)
+		}
+	}
+
 	// Schedule batch update for account last_used_at
 	s.deferredService.ScheduleLastUsedUpdate(account.ID)