Merge pull request #1027 from touwaeriol/feat/ignore-insufficient-balance-errors

feat(ops): add ignore insufficient balance errors toggle and extract error constants

Dockerfile.goreleaser

@@ -5,7 +5,12 @@
 # It only packages the pre-built binary, no compilation needed.
 # =============================================================================
 
-FROM alpine:3.19
+ARG ALPINE_IMAGE=alpine:3.21
+ARG POSTGRES_IMAGE=postgres:18-alpine
+
+FROM ${POSTGRES_IMAGE} AS pg-client
+
+FROM ${ALPINE_IMAGE}
 
 LABEL maintainer="Wei-Shaw <github.com/Wei-Shaw>"
 LABEL description="Sub2API - AI API Gateway Platform"
@@ -16,8 +21,20 @@ RUN apk add --no-cache \
     ca-certificates \
     tzdata \
     curl \
+    libpq \
+    zstd-libs \
+    lz4-libs \
+    krb5-libs \
+    libldap \
+    libedit \
     && rm -rf /var/cache/apk/*
 
+# Copy pg_dump and psql from a version-matched PostgreSQL image so backup and
+# restore work in the runtime container without requiring Docker socket access.
+COPY --from=pg-client /usr/local/bin/pg_dump /usr/local/bin/pg_dump
+COPY --from=pg-client /usr/local/bin/psql /usr/local/bin/psql
+COPY --from=pg-client /usr/local/lib/libpq.so.5* /usr/local/lib/
+
 # Create non-root user
 RUN addgroup -g 1000 sub2api && \
     adduser -u 1000 -G sub2api -s /bin/sh -D sub2api

backend/internal/handler/admin/group_handler.go

@@ -1,6 +1,9 @@
 package admin
 
 import (
+	"bytes"
+	"encoding/json"
+	"fmt"
 	"strconv"
 	"strings"
 
@@ -16,6 +19,55 @@ type GroupHandler struct {
 	adminService service.AdminService
 }
 
+type optionalLimitField struct {
+	set   bool
+	value *float64
+}
+
+func (f *optionalLimitField) UnmarshalJSON(data []byte) error {
+	f.set = true
+
+	trimmed := bytes.TrimSpace(data)
+	if bytes.Equal(trimmed, []byte("null")) {
+		f.value = nil
+		return nil
+	}
+
+	var number float64
+	if err := json.Unmarshal(trimmed, &number); err == nil {
+		f.value = &number
+		return nil
+	}
+
+	var text string
+	if err := json.Unmarshal(trimmed, &text); err == nil {
+		text = strings.TrimSpace(text)
+		if text == "" {
+			f.value = nil
+			return nil
+		}
+		number, err = strconv.ParseFloat(text, 64)
+		if err != nil {
+			return fmt.Errorf("invalid numeric limit value %q: %w", text, err)
+		}
+		f.value = &number
+		return nil
+	}
+
+	return fmt.Errorf("invalid limit value: %s", string(trimmed))
+}
+
+func (f optionalLimitField) ToServiceInput() *float64 {
+	if !f.set {
+		return nil
+	}
+	if f.value != nil {
+		return f.value
+	}
+	zero := 0.0
+	return &zero
+}
+
 // NewGroupHandler creates a new admin group handler
 func NewGroupHandler(adminService service.AdminService) *GroupHandler {
 	return &GroupHandler{
@@ -25,15 +77,15 @@ func NewGroupHandler(adminService service.AdminService) *GroupHandler {
 
 // CreateGroupRequest represents create group request
 type CreateGroupRequest struct {
-	Name             string   `json:"name" binding:"required"`
-	Description      string   `json:"description"`
-	Platform         string   `json:"platform" binding:"omitempty,oneof=anthropic openai gemini antigravity sora"`
-	RateMultiplier   float64  `json:"rate_multiplier"`
-	IsExclusive      bool     `json:"is_exclusive"`
-	SubscriptionType string   `json:"subscription_type" binding:"omitempty,oneof=standard subscription"`
-	DailyLimitUSD    *float64 `json:"daily_limit_usd"`
-	WeeklyLimitUSD   *float64 `json:"weekly_limit_usd"`
-	MonthlyLimitUSD  *float64 `json:"monthly_limit_usd"`
+	Name             string             `json:"name" binding:"required"`
+	Description      string             `json:"description"`
+	Platform         string             `json:"platform" binding:"omitempty,oneof=anthropic openai gemini antigravity sora"`
+	RateMultiplier   float64            `json:"rate_multiplier"`
+	IsExclusive      bool               `json:"is_exclusive"`
+	SubscriptionType string             `json:"subscription_type" binding:"omitempty,oneof=standard subscription"`
+	DailyLimitUSD    optionalLimitField `json:"daily_limit_usd"`
+	WeeklyLimitUSD   optionalLimitField `json:"weekly_limit_usd"`
+	MonthlyLimitUSD  optionalLimitField `json:"monthly_limit_usd"`
 	// 图片生成计费配置（antigravity 和 gemini 平台使用，负数表示清除配置）
 	ImagePrice1K                    *float64 `json:"image_price_1k"`
 	ImagePrice2K                    *float64 `json:"image_price_2k"`
@@ -62,16 +114,16 @@ type CreateGroupRequest struct {
 
 // UpdateGroupRequest represents update group request
 type UpdateGroupRequest struct {
-	Name             string   `json:"name"`
-	Description      string   `json:"description"`
-	Platform         string   `json:"platform" binding:"omitempty,oneof=anthropic openai gemini antigravity sora"`
-	RateMultiplier   *float64 `json:"rate_multiplier"`
-	IsExclusive      *bool    `json:"is_exclusive"`
-	Status           string   `json:"status" binding:"omitempty,oneof=active inactive"`
-	SubscriptionType string   `json:"subscription_type" binding:"omitempty,oneof=standard subscription"`
-	DailyLimitUSD    *float64 `json:"daily_limit_usd"`
-	WeeklyLimitUSD   *float64 `json:"weekly_limit_usd"`
-	MonthlyLimitUSD  *float64 `json:"monthly_limit_usd"`
+	Name             string             `json:"name"`
+	Description      string             `json:"description"`
+	Platform         string             `json:"platform" binding:"omitempty,oneof=anthropic openai gemini antigravity sora"`
+	RateMultiplier   *float64           `json:"rate_multiplier"`
+	IsExclusive      *bool              `json:"is_exclusive"`
+	Status           string             `json:"status" binding:"omitempty,oneof=active inactive"`
+	SubscriptionType string             `json:"subscription_type" binding:"omitempty,oneof=standard subscription"`
+	DailyLimitUSD    optionalLimitField `json:"daily_limit_usd"`
+	WeeklyLimitUSD   optionalLimitField `json:"weekly_limit_usd"`
+	MonthlyLimitUSD  optionalLimitField `json:"monthly_limit_usd"`
 	// 图片生成计费配置（antigravity 和 gemini 平台使用，负数表示清除配置）
 	ImagePrice1K                    *float64 `json:"image_price_1k"`
 	ImagePrice2K                    *float64 `json:"image_price_2k"`
@@ -191,9 +243,9 @@ func (h *GroupHandler) Create(c *gin.Context) {
 		RateMultiplier:                  req.RateMultiplier,
 		IsExclusive:                     req.IsExclusive,
 		SubscriptionType:                req.SubscriptionType,
-		DailyLimitUSD:                   req.DailyLimitUSD,
-		WeeklyLimitUSD:                  req.WeeklyLimitUSD,
-		MonthlyLimitUSD:                 req.MonthlyLimitUSD,
+		DailyLimitUSD:                   req.DailyLimitUSD.ToServiceInput(),
+		WeeklyLimitUSD:                  req.WeeklyLimitUSD.ToServiceInput(),
+		MonthlyLimitUSD:                 req.MonthlyLimitUSD.ToServiceInput(),
 		ImagePrice1K:                    req.ImagePrice1K,
 		ImagePrice2K:                    req.ImagePrice2K,
 		ImagePrice4K:                    req.ImagePrice4K,
@@ -244,9 +296,9 @@ func (h *GroupHandler) Update(c *gin.Context) {
 		IsExclusive:                     req.IsExclusive,
 		Status:                          req.Status,
 		SubscriptionType:                req.SubscriptionType,
-		DailyLimitUSD:                   req.DailyLimitUSD,
-		WeeklyLimitUSD:                  req.WeeklyLimitUSD,
-		MonthlyLimitUSD:                 req.MonthlyLimitUSD,
+		DailyLimitUSD:                   req.DailyLimitUSD.ToServiceInput(),
+		WeeklyLimitUSD:                  req.WeeklyLimitUSD.ToServiceInput(),
+		MonthlyLimitUSD:                 req.MonthlyLimitUSD.ToServiceInput(),
 		ImagePrice1K:                    req.ImagePrice1K,
 		ImagePrice2K:                    req.ImagePrice2K,
 		ImagePrice4K:                    req.ImagePrice4K,

backend/internal/handler/admin/setting_handler.go

@@ -80,6 +80,7 @@ func (h *SettingHandler) GetSettings(c *gin.Context) {
 		RegistrationEmailSuffixWhitelist:     settings.RegistrationEmailSuffixWhitelist,
 		PromoCodeEnabled:                     settings.PromoCodeEnabled,
 		PasswordResetEnabled:                 settings.PasswordResetEnabled,
+		FrontendURL:                          settings.FrontendURL,
 		InvitationCodeEnabled:                settings.InvitationCodeEnabled,
 		TotpEnabled:                          settings.TotpEnabled,
 		TotpEncryptionKeyConfigured:          h.settingService.IsTotpEncryptionKeyConfigured(),
@@ -137,6 +138,7 @@ type UpdateSettingsRequest struct {
 	RegistrationEmailSuffixWhitelist []string `json:"registration_email_suffix_whitelist"`
 	PromoCodeEnabled                 bool     `json:"promo_code_enabled"`
 	PasswordResetEnabled             bool     `json:"password_reset_enabled"`
+	FrontendURL                      string   `json:"frontend_url"`
 	InvitationCodeEnabled            bool     `json:"invitation_code_enabled"`
 	TotpEnabled                      bool     `json:"totp_enabled"` // TOTP 双因素认证
 
@@ -326,6 +328,15 @@ func (h *SettingHandler) UpdateSettings(c *gin.Context) {
 		}
 	}
 
+	// Frontend URL 验证
+	req.FrontendURL = strings.TrimSpace(req.FrontendURL)
+	if req.FrontendURL != "" {
+		if err := config.ValidateAbsoluteHTTPURL(req.FrontendURL); err != nil {
+			response.BadRequest(c, "Frontend URL must be an absolute http(s) URL")
+			return
+		}
+	}
+
 	// 自定义菜单项验证
 	const (
 		maxCustomMenuItems    = 20
@@ -437,6 +448,7 @@ func (h *SettingHandler) UpdateSettings(c *gin.Context) {
 		RegistrationEmailSuffixWhitelist: req.RegistrationEmailSuffixWhitelist,
 		PromoCodeEnabled:                 req.PromoCodeEnabled,
 		PasswordResetEnabled:             req.PasswordResetEnabled,
+		FrontendURL:                      req.FrontendURL,
 		InvitationCodeEnabled:            req.InvitationCodeEnabled,
 		TotpEnabled:                      req.TotpEnabled,
 		SMTPHost:                         req.SMTPHost,
@@ -531,6 +543,7 @@ func (h *SettingHandler) UpdateSettings(c *gin.Context) {
 		RegistrationEmailSuffixWhitelist:     updatedSettings.RegistrationEmailSuffixWhitelist,
 		PromoCodeEnabled:                     updatedSettings.PromoCodeEnabled,
 		PasswordResetEnabled:                 updatedSettings.PasswordResetEnabled,
+		FrontendURL:                          updatedSettings.FrontendURL,
 		InvitationCodeEnabled:                updatedSettings.InvitationCodeEnabled,
 		TotpEnabled:                          updatedSettings.TotpEnabled,
 		TotpEncryptionKeyConfigured:          h.settingService.IsTotpEncryptionKeyConfigured(),
@@ -614,6 +627,9 @@ func diffSettings(before *service.SystemSettings, after *service.SystemSettings,
 	if before.PasswordResetEnabled != after.PasswordResetEnabled {
 		changed = append(changed, "password_reset_enabled")
 	}
+	if before.FrontendURL != after.FrontendURL {
+		changed = append(changed, "frontend_url")
+	}
 	if before.TotpEnabled != after.TotpEnabled {
 		changed = append(changed, "totp_enabled")
 	}

backend/internal/handler/auth_handler.go

@@ -459,9 +459,9 @@ func (h *AuthHandler) ForgotPassword(c *gin.Context) {
 		return
 	}
 
-	frontendBaseURL := strings.TrimSpace(h.cfg.Server.FrontendURL)
+	frontendBaseURL := strings.TrimSpace(h.settingSvc.GetFrontendURL(c.Request.Context()))
 	if frontendBaseURL == "" {
-		slog.Error("server.frontend_url not configured; cannot build password reset link")
+		slog.Error("frontend_url not configured in settings or config; cannot build password reset link")
 		response.InternalError(c, "Password reset is not configured")
 		return
 	}

backend/internal/handler/dto/mappers.go

@@ -523,6 +523,8 @@ func usageLogFromServiceUser(l *service.UsageLog) UsageLog {
 		Model:                 l.Model,
 		ServiceTier:           l.ServiceTier,
 		ReasoningEffort:       l.ReasoningEffort,
+		InboundEndpoint:       l.InboundEndpoint,
+		UpstreamEndpoint:      l.UpstreamEndpoint,
 		GroupID:               l.GroupID,
 		SubscriptionID:        l.SubscriptionID,
 		InputTokens:           l.InputTokens,

backend/internal/handler/dto/mappers_usage_test.go

@@ -76,10 +76,14 @@ func TestUsageLogFromService_IncludesServiceTierForUserAndAdmin(t *testing.T) {
 	t.Parallel()
 
 	serviceTier := "priority"
+	inboundEndpoint := "/v1/chat/completions"
+	upstreamEndpoint := "/v1/responses"
 	log := &service.UsageLog{
 		RequestID:             "req_3",
 		Model:                 "gpt-5.4",
 		ServiceTier:           &serviceTier,
+		InboundEndpoint:       &inboundEndpoint,
+		UpstreamEndpoint:      &upstreamEndpoint,
 		AccountRateMultiplier: f64Ptr(1.5),
 	}
 
@@ -88,8 +92,16 @@ func TestUsageLogFromService_IncludesServiceTierForUserAndAdmin(t *testing.T) {
 
 	require.NotNil(t, userDTO.ServiceTier)
 	require.Equal(t, serviceTier, *userDTO.ServiceTier)
+	require.NotNil(t, userDTO.InboundEndpoint)
+	require.Equal(t, inboundEndpoint, *userDTO.InboundEndpoint)
+	require.NotNil(t, userDTO.UpstreamEndpoint)
+	require.Equal(t, upstreamEndpoint, *userDTO.UpstreamEndpoint)
 	require.NotNil(t, adminDTO.ServiceTier)
 	require.Equal(t, serviceTier, *adminDTO.ServiceTier)
+	require.NotNil(t, adminDTO.InboundEndpoint)
+	require.Equal(t, inboundEndpoint, *adminDTO.InboundEndpoint)
+	require.NotNil(t, adminDTO.UpstreamEndpoint)
+	require.Equal(t, upstreamEndpoint, *adminDTO.UpstreamEndpoint)
 	require.NotNil(t, adminDTO.AccountRateMultiplier)
 	require.InDelta(t, 1.5, *adminDTO.AccountRateMultiplier, 1e-12)
 }

backend/internal/handler/dto/settings.go

@@ -22,6 +22,7 @@ type SystemSettings struct {
 	RegistrationEmailSuffixWhitelist []string `json:"registration_email_suffix_whitelist"`
 	PromoCodeEnabled                 bool     `json:"promo_code_enabled"`
 	PasswordResetEnabled             bool     `json:"password_reset_enabled"`
+	FrontendURL                      string   `json:"frontend_url"`
 	InvitationCodeEnabled            bool     `json:"invitation_code_enabled"`
 	TotpEnabled                      bool     `json:"totp_enabled"`                   // TOTP 双因素认证
 	TotpEncryptionKeyConfigured      bool     `json:"totp_encryption_key_configured"` // TOTP 加密密钥是否已配置

backend/internal/handler/dto/types.go

@@ -334,9 +334,13 @@ type UsageLog struct {
 	Model     string `json:"model"`
 	// ServiceTier records the OpenAI service tier used for billing, e.g. "priority" / "flex".
 	ServiceTier *string `json:"service_tier,omitempty"`
-	// ReasoningEffort is the request's reasoning effort level (OpenAI Responses API).
-	// nil means not provided / not applicable.
+	// ReasoningEffort is the request's reasoning effort level.
+	// OpenAI: "low"/"medium"/"high"/"xhigh"; Claude: "low"/"medium"/"high"/"max".
 	ReasoningEffort *string `json:"reasoning_effort,omitempty"`
+	// InboundEndpoint is the client-facing API endpoint path, e.g. /v1/chat/completions.
+	InboundEndpoint *string `json:"inbound_endpoint,omitempty"`
+	// UpstreamEndpoint is the normalized upstream endpoint path, e.g. /v1/responses.
+	UpstreamEndpoint *string `json:"upstream_endpoint,omitempty"`
 
 	GroupID        *int64 `json:"group_id"`
 	SubscriptionID *int64 `json:"subscription_id"`

backend/internal/handler/gateway_handler.go

@@ -391,6 +391,8 @@ func (h *GatewayHandler) Messages(c *gin.Context) {
 			if fs.SwitchCount > 0 {
 				requestCtx = service.WithAccountSwitchCount(requestCtx, fs.SwitchCount, h.metadataBridgeEnabled())
 			}
+			// 记录 Forward 前已写入字节数，Forward 后若增加则说明 SSE 内容已发，禁止 failover
+			writerSizeBeforeForward := c.Writer.Size()
 			if account.Platform == service.PlatformAntigravity {
 				result, err = h.antigravityGatewayService.ForwardGemini(requestCtx, c, account, reqModel, "generateContent", reqStream, body, hasBoundSession)
 			} else {
@@ -402,6 +404,11 @@ func (h *GatewayHandler) Messages(c *gin.Context) {
 			if err != nil {
 				var failoverErr *service.UpstreamFailoverError
 				if errors.As(err, &failoverErr) {
+					// 流式内容已写入客户端，无法撤销，禁止 failover 以防止流拼接腐化
+					if c.Writer.Size() != writerSizeBeforeForward {
+						h.handleFailoverExhausted(c, failoverErr, service.PlatformGemini, true)
+						return
+					}
 					action := fs.HandleFailoverError(c.Request.Context(), h.gatewayService, account.ID, account.Platform, failoverErr)
 					switch action {
 					case FailoverContinue:
@@ -436,6 +443,10 @@ func (h *GatewayHandler) Messages(c *gin.Context) {
 			clientIP := ip.GetClientIP(c)
 			requestPayloadHash := service.HashUsageRequestPayload(body)
 
+			if result.ReasoningEffort == nil {
+				result.ReasoningEffort = service.NormalizeClaudeOutputEffort(parsedReq.OutputEffort)
+			}
+
 			// 使用量记录通过有界 worker 池提交，避免请求热路径创建无界 goroutine。
 			h.submitUsageRecordTask(func(ctx context.Context) {
 				if err := h.gatewayService.RecordUsage(ctx, &service.RecordUsageInput{
@@ -637,6 +648,8 @@ func (h *GatewayHandler) Messages(c *gin.Context) {
 			if fs.SwitchCount > 0 {
 				requestCtx = service.WithAccountSwitchCount(requestCtx, fs.SwitchCount, h.metadataBridgeEnabled())
 			}
+			// 记录 Forward 前已写入字节数，Forward 后若增加则说明 SSE 内容已发，禁止 failover
+			writerSizeBeforeForward := c.Writer.Size()
 			if account.Platform == service.PlatformAntigravity && account.Type != service.AccountTypeAPIKey {
 				result, err = h.antigravityGatewayService.Forward(requestCtx, c, account, body, hasBoundSession)
 			} else {
@@ -706,6 +719,11 @@ func (h *GatewayHandler) Messages(c *gin.Context) {
 				}
 				var failoverErr *service.UpstreamFailoverError
 				if errors.As(err, &failoverErr) {
+					// 流式内容已写入客户端，无法撤销，禁止 failover 以防止流拼接腐化
+					if c.Writer.Size() != writerSizeBeforeForward {
+						h.handleFailoverExhausted(c, failoverErr, account.Platform, true)
+						return
+					}
 					action := fs.HandleFailoverError(c.Request.Context(), h.gatewayService, account.ID, account.Platform, failoverErr)
 					switch action {
 					case FailoverContinue:
@@ -740,6 +758,10 @@ func (h *GatewayHandler) Messages(c *gin.Context) {
 			clientIP := ip.GetClientIP(c)
 			requestPayloadHash := service.HashUsageRequestPayload(body)
 
+			if result.ReasoningEffort == nil {
+				result.ReasoningEffort = service.NormalizeClaudeOutputEffort(parsedReq.OutputEffort)
+			}
+
 			// 使用量记录通过有界 worker 池提交，避免请求热路径创建无界 goroutine。
 			h.submitUsageRecordTask(func(ctx context.Context) {
 				if err := h.gatewayService.RecordUsage(ctx, &service.RecordUsageInput{

backend/internal/handler/gateway_handler_stream_failover_test.go

@@ -0,0 +1,122 @@
+package handler
+
+import (
+	"net/http"
+	"net/http/httptest"
+	"strings"
+	"testing"
+
+	"github.com/Wei-Shaw/sub2api/internal/service"
+	"github.com/gin-gonic/gin"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+// partialMessageStartSSE 模拟 handleStreamingResponse 已写入的首批 SSE 事件。
+const partialMessageStartSSE = "event: message_start\ndata: {\"type\":\"message_start\",\"message\":{\"id\":\"msg_01\",\"type\":\"message\",\"role\":\"assistant\",\"content\":[],\"model\":\"claude-sonnet-4-5\",\"stop_reason\":null,\"stop_sequence\":null,\"usage\":{\"input_tokens\":10,\"output_tokens\":1}}}\n\n" +
+	"event: content_block_start\ndata: {\"type\":\"content_block_start\",\"index\":0,\"content_block\":{\"type\":\"text\",\"text\":\"\"}}\n\n"
+
+// TestStreamWrittenGuard_MessagesPath_AbortFailoverOnSSEContentWritten 验证：
+// 当 Forward 在返回 UpstreamFailoverError 前已向客户端写入 SSE 内容时，
+// 故障转移保护逻辑必须终止循环并发送 SSE 错误事件，而不是进行下一次 Forward。
+// 具体验证：
+//  1. c.Writer.Size() 检测条件正确触发（字节数已增加）
+//  2. handleFailoverExhausted 以 streamStarted=true 调用后，响应体以 SSE 错误事件结尾
+//  3. 响应体中只出现一个 message_start，不存在第二个（防止流拼接腐化）
+func TestStreamWrittenGuard_MessagesPath_AbortFailoverOnSSEContentWritten(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	w := httptest.NewRecorder()
+	c, _ := gin.CreateTestContext(w)
+	c.Request = httptest.NewRequest(http.MethodPost, "/v1/messages", nil)
+
+	// 步骤 1：记录 Forward 前的 writer size（模拟 writerSizeBeforeForward := c.Writer.Size()）
+	sizeBeforeForward := c.Writer.Size()
+	require.Equal(t, -1, sizeBeforeForward, "gin writer 初始 Size 应为 -1（未写入任何字节）")
+
+	// 步骤 2：模拟 Forward 已向客户端写入部分 SSE 内容（message_start + content_block_start）
+	_, err := c.Writer.Write([]byte(partialMessageStartSSE))
+	require.NoError(t, err)
+
+	// 步骤 3：验证守卫条件成立（c.Writer.Size() != sizeBeforeForward）
+	require.NotEqual(t, sizeBeforeForward, c.Writer.Size(),
+		"写入 SSE 内容后 writer size 必须增加，守卫条件应为 true")
+
+	// 步骤 4：模拟 UpstreamFailoverError（上游在流中途返回 403）
+	failoverErr := &service.UpstreamFailoverError{
+		StatusCode:   http.StatusForbidden,
+		ResponseBody: []byte(`{"error":{"type":"permission_error","message":"forbidden"}}`),
+	}
+
+	// 步骤 5：守卫触发 → 调用 handleFailoverExhausted，streamStarted=true
+	h := &GatewayHandler{}
+	h.handleFailoverExhausted(c, failoverErr, service.PlatformAnthropic, true)
+
+	body := w.Body.String()
+
+	// 断言 A：响应体中包含最初写入的 message_start SSE 事件行
+	require.Contains(t, body, "event: message_start", "响应体应包含已写入的 message_start SSE 事件")
+
+	// 断言 B：响应体以 SSE 错误事件结尾（data: {"type":"error",...}\n\n）
+	require.True(t, strings.HasSuffix(strings.TrimRight(body, "\n"), "}"),
+		"响应体应以 JSON 对象结尾（SSE error event 的 data 字段）")
+	require.Contains(t, body, `"type":"error"`, "响应体末尾必须包含 SSE 错误事件")
+
+	// 断言 C：SSE event 行 "event: message_start" 只出现一次（防止双 message_start 拼接腐化）
+	firstIdx := strings.Index(body, "event: message_start")
+	lastIdx := strings.LastIndex(body, "event: message_start")
+	assert.Equal(t, firstIdx, lastIdx,
+		"响应体中 'event: message_start' 必须只出现一次，不得因 failover 拼接导致两次")
+}
+
+// TestStreamWrittenGuard_GeminiPath_AbortFailoverOnSSEContentWritten 与上述测试相同，
+// 验证 Gemini 路径使用 service.PlatformGemini（而非 account.Platform）时行为一致。
+func TestStreamWrittenGuard_GeminiPath_AbortFailoverOnSSEContentWritten(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	w := httptest.NewRecorder()
+	c, _ := gin.CreateTestContext(w)
+	c.Request = httptest.NewRequest(http.MethodPost, "/v1beta/models/gemini-2.0-flash:streamGenerateContent", nil)
+
+	sizeBeforeForward := c.Writer.Size()
+
+	_, err := c.Writer.Write([]byte(partialMessageStartSSE))
+	require.NoError(t, err)
+
+	require.NotEqual(t, sizeBeforeForward, c.Writer.Size())
+
+	failoverErr := &service.UpstreamFailoverError{
+		StatusCode: http.StatusForbidden,
+	}
+
+	h := &GatewayHandler{}
+	h.handleFailoverExhausted(c, failoverErr, service.PlatformGemini, true)
+
+	body := w.Body.String()
+
+	require.Contains(t, body, "event: message_start")
+	require.Contains(t, body, `"type":"error"`)
+
+	firstIdx := strings.Index(body, "event: message_start")
+	lastIdx := strings.LastIndex(body, "event: message_start")
+	assert.Equal(t, firstIdx, lastIdx, "Gemini 路径不得出现双 message_start")
+}
+
+// TestStreamWrittenGuard_NoByteWritten_GuardNotTriggered 验证反向场景：
+// 当 Forward 返回 UpstreamFailoverError 时若未向客户端写入任何 SSE 内容，
+// 守卫条件（c.Writer.Size() != sizeBeforeForward）为 false，不应中止 failover。
+func TestStreamWrittenGuard_NoByteWritten_GuardNotTriggered(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	w := httptest.NewRecorder()
+	c, _ := gin.CreateTestContext(w)
+	c.Request = httptest.NewRequest(http.MethodPost, "/v1/messages", nil)
+
+	// 模拟 writerSizeBeforeForward：初始为 -1
+	sizeBeforeForward := c.Writer.Size()
+
+	// Forward 未写入任何字节直接返回错误（例如 401 发生在连接建立前）
+	// c.Writer.Size() 仍为 -1
+
+	// 守卫条件：sizeBeforeForward == c.Writer.Size() → 不触发
+	guardTriggered := c.Writer.Size() != sizeBeforeForward
+	require.False(t, guardTriggered,
+		"未写入任何字节时，守卫条件必须为 false，应允许正常 failover 继续")
+}

backend/internal/handler/openai_chat_completions.go

@@ -256,14 +256,16 @@ func (h *OpenAIGatewayHandler) ChatCompletions(c *gin.Context) {
 
 		h.submitUsageRecordTask(func(ctx context.Context) {
 			if err := h.gatewayService.RecordUsage(ctx, &service.OpenAIRecordUsageInput{
-				Result:        result,
-				APIKey:        apiKey,
-				User:          apiKey.User,
-				Account:       account,
-				Subscription:  subscription,
-				UserAgent:     userAgent,
-				IPAddress:     clientIP,
-				APIKeyService: h.apiKeyService,
+				Result:           result,
+				APIKey:           apiKey,
+				User:             apiKey.User,
+				Account:          account,
+				Subscription:     subscription,
+				InboundEndpoint:  normalizedOpenAIInboundEndpoint(c, openAIInboundEndpointChatCompletions),
+				UpstreamEndpoint: normalizedOpenAIUpstreamEndpoint(c, openAIUpstreamEndpointResponses),
+				UserAgent:        userAgent,
+				IPAddress:        clientIP,
+				APIKeyService:    h.apiKeyService,
 			}); err != nil {
 				logger.L().With(
 					zap.String("component", "handler.openai_gateway.chat_completions"),

backend/internal/handler/openai_gateway_endpoint_normalization_test.go

@@ -0,0 +1,57 @@
+package handler
+
+import (
+	"net/http"
+	"net/http/httptest"
+	"testing"
+
+	"github.com/gin-gonic/gin"
+	"github.com/stretchr/testify/require"
+)
+
+func TestNormalizedOpenAIUpstreamEndpoint(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+
+	tests := []struct {
+		name     string
+		path     string
+		fallback string
+		want     string
+	}{
+		{
+			name:     "responses root maps to responses upstream",
+			path:     "/v1/responses",
+			fallback: openAIUpstreamEndpointResponses,
+			want:     "/v1/responses",
+		},
+		{
+			name:     "responses compact keeps compact suffix",
+			path:     "/openai/v1/responses/compact",
+			fallback: openAIUpstreamEndpointResponses,
+			want:     "/v1/responses/compact",
+		},
+		{
+			name:     "responses nested suffix preserved",
+			path:     "/openai/v1/responses/compact/detail",
+			fallback: openAIUpstreamEndpointResponses,
+			want:     "/v1/responses/compact/detail",
+		},
+		{
+			name:     "non responses path uses fallback",
+			path:     "/v1/messages",
+			fallback: openAIUpstreamEndpointResponses,
+			want:     "/v1/responses",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			rec := httptest.NewRecorder()
+			c, _ := gin.CreateTestContext(rec)
+			c.Request = httptest.NewRequest(http.MethodPost, tt.path, nil)
+
+			got := normalizedOpenAIUpstreamEndpoint(c, tt.fallback)
+			require.Equal(t, tt.want, got)
+		})
+	}
+}

backend/internal/handler/openai_gateway_handler.go

@@ -37,6 +37,13 @@ type OpenAIGatewayHandler struct {
 	cfg                     *config.Config
 }
 
+const (
+	openAIInboundEndpointResponses       = "/v1/responses"
+	openAIInboundEndpointMessages        = "/v1/messages"
+	openAIInboundEndpointChatCompletions = "/v1/chat/completions"
+	openAIUpstreamEndpointResponses      = "/v1/responses"
+)
+
 // NewOpenAIGatewayHandler creates a new OpenAIGatewayHandler
 func NewOpenAIGatewayHandler(
 	gatewayService *service.OpenAIGatewayService,
@@ -362,6 +369,8 @@ func (h *OpenAIGatewayHandler) Responses(c *gin.Context) {
 				User:               apiKey.User,
 				Account:            account,
 				Subscription:       subscription,
+				InboundEndpoint:    normalizedOpenAIInboundEndpoint(c, openAIInboundEndpointResponses),
+				UpstreamEndpoint:   normalizedOpenAIUpstreamEndpoint(c, openAIUpstreamEndpointResponses),
 				UserAgent:          userAgent,
 				IPAddress:          clientIP,
 				RequestPayloadHash: requestPayloadHash,
@@ -738,6 +747,8 @@ func (h *OpenAIGatewayHandler) Messages(c *gin.Context) {
 				User:               apiKey.User,
 				Account:            account,
 				Subscription:       subscription,
+				InboundEndpoint:    normalizedOpenAIInboundEndpoint(c, openAIInboundEndpointMessages),
+				UpstreamEndpoint:   normalizedOpenAIUpstreamEndpoint(c, openAIUpstreamEndpointResponses),
 				UserAgent:          userAgent,
 				IPAddress:          clientIP,
 				RequestPayloadHash: requestPayloadHash,
@@ -1235,6 +1246,8 @@ func (h *OpenAIGatewayHandler) ResponsesWebSocket(c *gin.Context) {
 					User:               apiKey.User,
 					Account:            account,
 					Subscription:       subscription,
+					InboundEndpoint:    normalizedOpenAIInboundEndpoint(c, openAIInboundEndpointResponses),
+					UpstreamEndpoint:   normalizedOpenAIUpstreamEndpoint(c, openAIUpstreamEndpointResponses),
 					UserAgent:          userAgent,
 					IPAddress:          clientIP,
 					RequestPayloadHash: service.HashUsageRequestPayload(firstMessage),
@@ -1530,6 +1543,62 @@ func openAIWSIngressFallbackSessionSeed(userID, apiKeyID int64, groupID *int64)
 	return fmt.Sprintf("openai_ws_ingress:%d:%d:%d", gid, userID, apiKeyID)
 }
 
+func normalizedOpenAIInboundEndpoint(c *gin.Context, fallback string) string {
+	path := strings.TrimSpace(fallback)
+	if c != nil {
+		if fullPath := strings.TrimSpace(c.FullPath()); fullPath != "" {
+			path = fullPath
+		} else if c.Request != nil && c.Request.URL != nil {
+			if requestPath := strings.TrimSpace(c.Request.URL.Path); requestPath != "" {
+				path = requestPath
+			}
+		}
+	}
+
+	switch {
+	case strings.Contains(path, openAIInboundEndpointChatCompletions):
+		return openAIInboundEndpointChatCompletions
+	case strings.Contains(path, openAIInboundEndpointMessages):
+		return openAIInboundEndpointMessages
+	case strings.Contains(path, openAIInboundEndpointResponses):
+		return openAIInboundEndpointResponses
+	default:
+		return path
+	}
+}
+
+func normalizedOpenAIUpstreamEndpoint(c *gin.Context, fallback string) string {
+	base := strings.TrimSpace(fallback)
+	if base == "" {
+		base = openAIUpstreamEndpointResponses
+	}
+	base = strings.TrimRight(base, "/")
+
+	if c == nil || c.Request == nil || c.Request.URL == nil {
+		return base
+	}
+
+	path := strings.TrimRight(strings.TrimSpace(c.Request.URL.Path), "/")
+	if path == "" {
+		return base
+	}
+
+	idx := strings.LastIndex(path, "/responses")
+	if idx < 0 {
+		return base
+	}
+
+	suffix := strings.TrimSpace(path[idx+len("/responses"):])
+	if suffix == "" || suffix == "/" {
+		return base
+	}
+	if !strings.HasPrefix(suffix, "/") {
+		return base
+	}
+
+	return base + suffix
+}
+
 func isOpenAIWSUpgradeRequest(r *http.Request) bool {
 	if r == nil {
 		return false

backend/internal/handler/ops_error_logger.go

@@ -26,6 +26,22 @@ const (
 	opsStreamKey      = "ops_stream"
 	opsRequestBodyKey = "ops_request_body"
 	opsAccountIDKey   = "ops_account_id"
+
+	// 错误过滤匹配常量 — shouldSkipOpsErrorLog 和错误分类共用
+	opsErrContextCanceled            = "context canceled"
+	opsErrNoAvailableAccounts        = "no available accounts"
+	opsErrInvalidAPIKey              = "invalid_api_key"
+	opsErrAPIKeyRequired             = "api_key_required"
+	opsErrInsufficientBalance        = "insufficient balance"
+	opsErrInsufficientAccountBalance = "insufficient account balance"
+	opsErrInsufficientQuota          = "insufficient_quota"
+
+	// 上游错误码常量 — 错误分类 (normalizeOpsErrorType / classifyOpsPhase / classifyOpsIsBusinessLimited)
+	opsCodeInsufficientBalance  = "INSUFFICIENT_BALANCE"
+	opsCodeUsageLimitExceeded   = "USAGE_LIMIT_EXCEEDED"
+	opsCodeSubscriptionNotFound = "SUBSCRIPTION_NOT_FOUND"
+	opsCodeSubscriptionInvalid  = "SUBSCRIPTION_INVALID"
+	opsCodeUserInactive         = "USER_INACTIVE"
 )
 
 const (
@@ -1024,9 +1040,9 @@ func normalizeOpsErrorType(errType string, code string) string {
 		return errType
 	}
 	switch strings.TrimSpace(code) {
-	case "INSUFFICIENT_BALANCE":
+	case opsCodeInsufficientBalance:
 		return "billing_error"
-	case "USAGE_LIMIT_EXCEEDED", "SUBSCRIPTION_NOT_FOUND", "SUBSCRIPTION_INVALID":
+	case opsCodeUsageLimitExceeded, opsCodeSubscriptionNotFound, opsCodeSubscriptionInvalid:
 		return "subscription_error"
 	default:
 		return "api_error"
@@ -1038,7 +1054,7 @@ func classifyOpsPhase(errType, message, code string) string {
 	// Standardized phases: request|auth|routing|upstream|network|internal
 	// Map billing/concurrency/response => request; scheduling => routing.
 	switch strings.TrimSpace(code) {
-	case "INSUFFICIENT_BALANCE", "USAGE_LIMIT_EXCEEDED", "SUBSCRIPTION_NOT_FOUND", "SUBSCRIPTION_INVALID":
+	case opsCodeInsufficientBalance, opsCodeUsageLimitExceeded, opsCodeSubscriptionNotFound, opsCodeSubscriptionInvalid:
 		return "request"
 	}
 
@@ -1057,7 +1073,7 @@ func classifyOpsPhase(errType, message, code string) string {
 	case "upstream_error", "overloaded_error":
 		return "upstream"
 	case "api_error":
-		if strings.Contains(msg, "no available accounts") {
+		if strings.Contains(msg, opsErrNoAvailableAccounts) {
 			return "routing"
 		}
 		return "internal"
@@ -1103,7 +1119,7 @@ func classifyOpsIsRetryable(errType string, statusCode int) bool {
 
 func classifyOpsIsBusinessLimited(errType, phase, code string, status int, message string) bool {
 	switch strings.TrimSpace(code) {
-	case "INSUFFICIENT_BALANCE", "USAGE_LIMIT_EXCEEDED", "SUBSCRIPTION_NOT_FOUND", "SUBSCRIPTION_INVALID", "USER_INACTIVE":
+	case opsCodeInsufficientBalance, opsCodeUsageLimitExceeded, opsCodeSubscriptionNotFound, opsCodeSubscriptionInvalid, opsCodeUserInactive:
 		return true
 	}
 	if phase == "billing" || phase == "concurrency" {
@@ -1197,21 +1213,30 @@ func shouldSkipOpsErrorLog(ctx context.Context, ops *service.OpsService, message
 
 	// Check if context canceled errors should be ignored (client disconnects)
 	if settings.IgnoreContextCanceled {
-		if strings.Contains(msgLower, "context canceled") || strings.Contains(bodyLower, "context canceled") {
+		if strings.Contains(msgLower, opsErrContextCanceled) || strings.Contains(bodyLower, opsErrContextCanceled) {
 			return true
 		}
 	}
 
 	// Check if "no available accounts" errors should be ignored
 	if settings.IgnoreNoAvailableAccounts {
-		if strings.Contains(msgLower, "no available accounts") || strings.Contains(bodyLower, "no available accounts") {
+		if strings.Contains(msgLower, opsErrNoAvailableAccounts) || strings.Contains(bodyLower, opsErrNoAvailableAccounts) {
 			return true
 		}
 	}
 
 	// Check if invalid/missing API key errors should be ignored (user misconfiguration)
 	if settings.IgnoreInvalidApiKeyErrors {
-		if strings.Contains(bodyLower, "invalid_api_key") || strings.Contains(bodyLower, "api_key_required") {
+		if strings.Contains(bodyLower, opsErrInvalidAPIKey) || strings.Contains(bodyLower, opsErrAPIKeyRequired) {
+			return true
+		}
+	}
+
+	// Check if insufficient balance errors should be ignored
+	if settings.IgnoreInsufficientBalanceErrors {
+		if strings.Contains(bodyLower, opsErrInsufficientBalance) || strings.Contains(bodyLower, opsErrInsufficientAccountBalance) ||
+			strings.Contains(bodyLower, opsErrInsufficientQuota) ||
+			strings.Contains(msgLower, opsErrInsufficientBalance) || strings.Contains(msgLower, opsErrInsufficientAccountBalance) {
 			return true
 		}
 	}

backend/internal/handler/sora_gateway_handler_test.go

@@ -334,6 +334,14 @@ func (s *stubUsageLogRepo) GetUsageTrendWithFilters(ctx context.Context, startTi
 func (s *stubUsageLogRepo) GetModelStatsWithFilters(ctx context.Context, startTime, endTime time.Time, userID, apiKeyID, accountID, groupID int64, requestType *int16, stream *bool, billingType *int8) ([]usagestats.ModelStat, error) {
 	return nil, nil
 }
+
+func (s *stubUsageLogRepo) GetEndpointStatsWithFilters(ctx context.Context, startTime, endTime time.Time, userID, apiKeyID, accountID, groupID int64, model string, requestType *int16, stream *bool, billingType *int8) ([]usagestats.EndpointStat, error) {
+	return []usagestats.EndpointStat{}, nil
+}
+
+func (s *stubUsageLogRepo) GetUpstreamEndpointStatsWithFilters(ctx context.Context, startTime, endTime time.Time, userID, apiKeyID, accountID, groupID int64, model string, requestType *int16, stream *bool, billingType *int8) ([]usagestats.EndpointStat, error) {
+	return []usagestats.EndpointStat{}, nil
+}
 func (s *stubUsageLogRepo) GetGroupStatsWithFilters(ctx context.Context, startTime, endTime time.Time, userID, apiKeyID, accountID, groupID int64, requestType *int16, stream *bool, billingType *int8) ([]usagestats.GroupStat, error) {
 	return nil, nil
 }

backend/internal/pkg/usagestats/usage_log_types.go

@@ -81,6 +81,15 @@ type ModelStat struct {
 	ActualCost          float64 `json:"actual_cost"` // 实际扣除
 }
 
+// EndpointStat represents usage statistics for a single request endpoint.
+type EndpointStat struct {
+	Endpoint    string  `json:"endpoint"`
+	Requests    int64   `json:"requests"`
+	TotalTokens int64   `json:"total_tokens"`
+	Cost        float64 `json:"cost"`        // 标准计费
+	ActualCost  float64 `json:"actual_cost"` // 实际扣除
+}
+
 // GroupStat represents usage statistics for a single group
 type GroupStat struct {
 	GroupID     int64   `json:"group_id"`
@@ -179,15 +188,18 @@ type UsageLogFilters struct {
 
 // UsageStats represents usage statistics
 type UsageStats struct {
-	TotalRequests     int64    `json:"total_requests"`
-	TotalInputTokens  int64    `json:"total_input_tokens"`
-	TotalOutputTokens int64    `json:"total_output_tokens"`
-	TotalCacheTokens  int64    `json:"total_cache_tokens"`
-	TotalTokens       int64    `json:"total_tokens"`
-	TotalCost         float64  `json:"total_cost"`
-	TotalActualCost   float64  `json:"total_actual_cost"`
-	TotalAccountCost  *float64 `json:"total_account_cost,omitempty"`
-	AverageDurationMs float64  `json:"average_duration_ms"`
+	TotalRequests     int64          `json:"total_requests"`
+	TotalInputTokens  int64          `json:"total_input_tokens"`
+	TotalOutputTokens int64          `json:"total_output_tokens"`
+	TotalCacheTokens  int64          `json:"total_cache_tokens"`
+	TotalTokens       int64          `json:"total_tokens"`
+	TotalCost         float64        `json:"total_cost"`
+	TotalActualCost   float64        `json:"total_actual_cost"`
+	TotalAccountCost  *float64       `json:"total_account_cost,omitempty"`
+	AverageDurationMs float64        `json:"average_duration_ms"`
+	Endpoints         []EndpointStat `json:"endpoints,omitempty"`
+	UpstreamEndpoints []EndpointStat `json:"upstream_endpoints,omitempty"`
+	EndpointPaths     []EndpointStat `json:"endpoint_paths,omitempty"`
 }
 
 // BatchUserUsageStats represents usage stats for a single user
@@ -254,7 +266,9 @@ type AccountUsageSummary struct {
 
 // AccountUsageStatsResponse represents the full usage statistics response for an account
 type AccountUsageStatsResponse struct {
-	History []AccountUsageHistory `json:"history"`
-	Summary AccountUsageSummary   `json:"summary"`
-	Models  []ModelStat           `json:"models"`
+	History           []AccountUsageHistory `json:"history"`
+	Summary           AccountUsageSummary   `json:"summary"`
+	Models            []ModelStat           `json:"models"`
+	Endpoints         []EndpointStat        `json:"endpoints"`
+	UpstreamEndpoints []EndpointStat        `json:"upstream_endpoints"`
 }

backend/internal/repository/usage_billing_repo.go

@@ -132,7 +132,7 @@ func (r *usageBillingRepository) applyUsageBillingEffects(ctx context.Context, t
 		}
 	}
 
-	if cmd.AccountQuotaCost > 0 && strings.EqualFold(cmd.AccountType, service.AccountTypeAPIKey) {
+	if cmd.AccountQuotaCost > 0 && (strings.EqualFold(cmd.AccountType, service.AccountTypeAPIKey) || strings.EqualFold(cmd.AccountType, service.AccountTypeBedrock)) {
 		if err := incrementUsageBillingAccountQuota(ctx, tx, cmd.AccountID, cmd.AccountQuotaCost); err != nil {
 			return err
 		}

backend/internal/repository/usage_log_repo.go

@@ -28,7 +28,7 @@ import (
 	gocache "github.com/patrickmn/go-cache"
 )
 
-const usageLogSelectColumns = "id, user_id, api_key_id, account_id, request_id, model, group_id, subscription_id, input_tokens, output_tokens, cache_creation_tokens, cache_read_tokens, cache_creation_5m_tokens, cache_creation_1h_tokens, input_cost, output_cost, cache_creation_cost, cache_read_cost, total_cost, actual_cost, rate_multiplier, account_rate_multiplier, billing_type, request_type, stream, openai_ws_mode, duration_ms, first_token_ms, user_agent, ip_address, image_count, image_size, media_type, service_tier, reasoning_effort, cache_ttl_overridden, created_at"
+const usageLogSelectColumns = "id, user_id, api_key_id, account_id, request_id, model, group_id, subscription_id, input_tokens, output_tokens, cache_creation_tokens, cache_read_tokens, cache_creation_5m_tokens, cache_creation_1h_tokens, input_cost, output_cost, cache_creation_cost, cache_read_cost, total_cost, actual_cost, rate_multiplier, account_rate_multiplier, billing_type, request_type, stream, openai_ws_mode, duration_ms, first_token_ms, user_agent, ip_address, image_count, image_size, media_type, service_tier, reasoning_effort, inbound_endpoint, upstream_endpoint, cache_ttl_overridden, created_at"
 
 var usageLogInsertArgTypes = [...]string{
 	"bigint",
@@ -65,6 +65,8 @@ var usageLogInsertArgTypes = [...]string{
 	"text",
 	"text",
 	"text",
+	"text",
+	"text",
 	"boolean",
 	"timestamptz",
 }
@@ -304,6 +306,8 @@ func (r *usageLogRepository) createSingle(ctx context.Context, sqlq sqlExecutor,
 			media_type,
 			service_tier,
 			reasoning_effort,
+			inbound_endpoint,
+			upstream_endpoint,
 			cache_ttl_overridden,
 			created_at
 		) VALUES (
@@ -312,7 +316,7 @@ func (r *usageLogRepository) createSingle(ctx context.Context, sqlq sqlExecutor,
 			$8, $9, $10, $11,
 			$12, $13,
 			$14, $15, $16, $17, $18, $19,
-			$20, $21, $22, $23, $24, $25, $26, $27, $28, $29, $30, $31, $32, $33, $34, $35, $36
+			$20, $21, $22, $23, $24, $25, $26, $27, $28, $29, $30, $31, $32, $33, $34, $35, $36, $37, $38
 		)
 		ON CONFLICT (request_id, api_key_id) DO NOTHING
 		RETURNING id, created_at
@@ -732,11 +736,13 @@ func buildUsageLogBatchInsertQuery(keys []string, preparedByKey map[string]usage
 			media_type,
 			service_tier,
 			reasoning_effort,
+			inbound_endpoint,
+			upstream_endpoint,
 			cache_ttl_overridden,
 			created_at
 		) AS (VALUES `)
 
-	args := make([]any, 0, len(keys)*37)
+	args := make([]any, 0, len(keys)*38)
 	argPos := 1
 	for idx, key := range keys {
 		if idx > 0 {
@@ -799,6 +805,8 @@ func buildUsageLogBatchInsertQuery(keys []string, preparedByKey map[string]usage
 				media_type,
 				service_tier,
 				reasoning_effort,
+				inbound_endpoint,
+				upstream_endpoint,
 				cache_ttl_overridden,
 				created_at
 			)
@@ -837,6 +845,8 @@ func buildUsageLogBatchInsertQuery(keys []string, preparedByKey map[string]usage
 				media_type,
 				service_tier,
 				reasoning_effort,
+				inbound_endpoint,
+				upstream_endpoint,
 				cache_ttl_overridden,
 				created_at
 			FROM input
@@ -915,11 +925,13 @@ func buildUsageLogBestEffortInsertQuery(preparedList []usageLogInsertPrepared) (
 			media_type,
 			service_tier,
 			reasoning_effort,
+			inbound_endpoint,
+			upstream_endpoint,
 			cache_ttl_overridden,
 			created_at
 		) AS (VALUES `)
 
-	args := make([]any, 0, len(preparedList)*36)
+	args := make([]any, 0, len(preparedList)*38)
 	argPos := 1
 	for idx, prepared := range preparedList {
 		if idx > 0 {
@@ -979,6 +991,8 @@ func buildUsageLogBestEffortInsertQuery(preparedList []usageLogInsertPrepared) (
 			media_type,
 			service_tier,
 			reasoning_effort,
+			inbound_endpoint,
+			upstream_endpoint,
 			cache_ttl_overridden,
 			created_at
 		)
@@ -1017,6 +1031,8 @@ func buildUsageLogBestEffortInsertQuery(preparedList []usageLogInsertPrepared) (
 			media_type,
 			service_tier,
 			reasoning_effort,
+			inbound_endpoint,
+			upstream_endpoint,
 			cache_ttl_overridden,
 			created_at
 		FROM input
@@ -1063,6 +1079,8 @@ func execUsageLogInsertNoResult(ctx context.Context, sqlq sqlExecutor, prepared
 			media_type,
 			service_tier,
 			reasoning_effort,
+			inbound_endpoint,
+			upstream_endpoint,
 			cache_ttl_overridden,
 			created_at
 		) VALUES (
@@ -1071,7 +1089,7 @@ func execUsageLogInsertNoResult(ctx context.Context, sqlq sqlExecutor, prepared
 			$8, $9, $10, $11,
 			$12, $13,
 			$14, $15, $16, $17, $18, $19,
-			$20, $21, $22, $23, $24, $25, $26, $27, $28, $29, $30, $31, $32, $33, $34, $35, $36
+			$20, $21, $22, $23, $24, $25, $26, $27, $28, $29, $30, $31, $32, $33, $34, $35, $36, $37, $38
 		)
 		ON CONFLICT (request_id, api_key_id) DO NOTHING
 	`, prepared.args...)
@@ -1101,6 +1119,8 @@ func prepareUsageLogInsert(log *service.UsageLog) usageLogInsertPrepared {
 	mediaType := nullString(log.MediaType)
 	serviceTier := nullString(log.ServiceTier)
 	reasoningEffort := nullString(log.ReasoningEffort)
+	inboundEndpoint := nullString(log.InboundEndpoint)
+	upstreamEndpoint := nullString(log.UpstreamEndpoint)
 
 	var requestIDArg any
 	if requestID != "" {
@@ -1147,6 +1167,8 @@ func prepareUsageLogInsert(log *service.UsageLog) usageLogInsertPrepared {
 			mediaType,
 			serviceTier,
 			reasoningEffort,
+			inboundEndpoint,
+			upstreamEndpoint,
 			log.CacheTTLOverridden,
 			createdAt,
 		},
@@ -2505,7 +2527,7 @@ func (r *usageLogRepository) ListWithFilters(ctx context.Context, params paginat
 		args = append(args, *filters.StartTime)
 	}
 	if filters.EndTime != nil {
-		conditions = append(conditions, fmt.Sprintf("created_at <= $%d", len(args)+1))
+		conditions = append(conditions, fmt.Sprintf("created_at < $%d", len(args)+1))
 		args = append(args, *filters.EndTime)
 	}
 
@@ -3040,7 +3062,7 @@ func (r *usageLogRepository) GetStatsWithFilters(ctx context.Context, filters Us
 		args = append(args, *filters.StartTime)
 	}
 	if filters.EndTime != nil {
-		conditions = append(conditions, fmt.Sprintf("created_at <= $%d", len(args)+1))
+		conditions = append(conditions, fmt.Sprintf("created_at < $%d", len(args)+1))
 		args = append(args, *filters.EndTime)
 	}
 
@@ -3080,6 +3102,35 @@ func (r *usageLogRepository) GetStatsWithFilters(ctx context.Context, filters Us
 		stats.TotalAccountCost = &totalAccountCost
 	}
 	stats.TotalTokens = stats.TotalInputTokens + stats.TotalOutputTokens + stats.TotalCacheTokens
+
+	start := time.Unix(0, 0).UTC()
+	if filters.StartTime != nil {
+		start = *filters.StartTime
+	}
+	end := time.Now().UTC()
+	if filters.EndTime != nil {
+		end = *filters.EndTime
+	}
+
+	endpoints, endpointErr := r.GetEndpointStatsWithFilters(ctx, start, end, filters.UserID, filters.APIKeyID, filters.AccountID, filters.GroupID, filters.Model, filters.RequestType, filters.Stream, filters.BillingType)
+	if endpointErr != nil {
+		logger.LegacyPrintf("repository.usage_log", "GetEndpointStatsWithFilters failed in GetStatsWithFilters: %v", endpointErr)
+		endpoints = []EndpointStat{}
+	}
+	upstreamEndpoints, upstreamEndpointErr := r.GetUpstreamEndpointStatsWithFilters(ctx, start, end, filters.UserID, filters.APIKeyID, filters.AccountID, filters.GroupID, filters.Model, filters.RequestType, filters.Stream, filters.BillingType)
+	if upstreamEndpointErr != nil {
+		logger.LegacyPrintf("repository.usage_log", "GetUpstreamEndpointStatsWithFilters failed in GetStatsWithFilters: %v", upstreamEndpointErr)
+		upstreamEndpoints = []EndpointStat{}
+	}
+	endpointPaths, endpointPathErr := r.getEndpointPathStatsWithFilters(ctx, start, end, filters.UserID, filters.APIKeyID, filters.AccountID, filters.GroupID, filters.Model, filters.RequestType, filters.Stream, filters.BillingType)
+	if endpointPathErr != nil {
+		logger.LegacyPrintf("repository.usage_log", "getEndpointPathStatsWithFilters failed in GetStatsWithFilters: %v", endpointPathErr)
+		endpointPaths = []EndpointStat{}
+	}
+	stats.Endpoints = endpoints
+	stats.UpstreamEndpoints = upstreamEndpoints
+	stats.EndpointPaths = endpointPaths
+
 	return stats, nil
 }
 
@@ -3092,6 +3143,163 @@ type AccountUsageSummary = usagestats.AccountUsageSummary
 // AccountUsageStatsResponse represents the full usage statistics response for an account
 type AccountUsageStatsResponse = usagestats.AccountUsageStatsResponse
 
+// EndpointStat represents endpoint usage statistics row.
+type EndpointStat = usagestats.EndpointStat
+
+func (r *usageLogRepository) getEndpointStatsByColumnWithFilters(ctx context.Context, endpointColumn string, startTime, endTime time.Time, userID, apiKeyID, accountID, groupID int64, model string, requestType *int16, stream *bool, billingType *int8) (results []EndpointStat, err error) {
+	actualCostExpr := "COALESCE(SUM(actual_cost), 0) as actual_cost"
+	if accountID > 0 && userID == 0 && apiKeyID == 0 {
+		actualCostExpr = "COALESCE(SUM(total_cost * COALESCE(account_rate_multiplier, 1)), 0) as actual_cost"
+	}
+
+	query := fmt.Sprintf(`
+		SELECT
+			COALESCE(NULLIF(TRIM(%s), ''), 'unknown') AS endpoint,
+			COUNT(*) AS requests,
+			COALESCE(SUM(input_tokens + output_tokens + cache_creation_tokens + cache_read_tokens), 0) AS total_tokens,
+			COALESCE(SUM(total_cost), 0) as cost,
+			%s
+		FROM usage_logs
+		WHERE created_at >= $1 AND created_at < $2
+	`, endpointColumn, actualCostExpr)
+
+	args := []any{startTime, endTime}
+	if userID > 0 {
+		query += fmt.Sprintf(" AND user_id = $%d", len(args)+1)
+		args = append(args, userID)
+	}
+	if apiKeyID > 0 {
+		query += fmt.Sprintf(" AND api_key_id = $%d", len(args)+1)
+		args = append(args, apiKeyID)
+	}
+	if accountID > 0 {
+		query += fmt.Sprintf(" AND account_id = $%d", len(args)+1)
+		args = append(args, accountID)
+	}
+	if groupID > 0 {
+		query += fmt.Sprintf(" AND group_id = $%d", len(args)+1)
+		args = append(args, groupID)
+	}
+	if model != "" {
+		query += fmt.Sprintf(" AND model = $%d", len(args)+1)
+		args = append(args, model)
+	}
+	query, args = appendRequestTypeOrStreamQueryFilter(query, args, requestType, stream)
+	if billingType != nil {
+		query += fmt.Sprintf(" AND billing_type = $%d", len(args)+1)
+		args = append(args, int16(*billingType))
+	}
+	query += " GROUP BY endpoint ORDER BY requests DESC"
+
+	rows, err := r.sql.QueryContext(ctx, query, args...)
+	if err != nil {
+		return nil, err
+	}
+	defer func() {
+		if closeErr := rows.Close(); closeErr != nil && err == nil {
+			err = closeErr
+			results = nil
+		}
+	}()
+
+	results = make([]EndpointStat, 0)
+	for rows.Next() {
+		var row EndpointStat
+		if err := rows.Scan(&row.Endpoint, &row.Requests, &row.TotalTokens, &row.Cost, &row.ActualCost); err != nil {
+			return nil, err
+		}
+		results = append(results, row)
+	}
+	if err := rows.Err(); err != nil {
+		return nil, err
+	}
+	return results, nil
+}
+
+func (r *usageLogRepository) getEndpointPathStatsWithFilters(ctx context.Context, startTime, endTime time.Time, userID, apiKeyID, accountID, groupID int64, model string, requestType *int16, stream *bool, billingType *int8) (results []EndpointStat, err error) {
+	actualCostExpr := "COALESCE(SUM(actual_cost), 0) as actual_cost"
+	if accountID > 0 && userID == 0 && apiKeyID == 0 {
+		actualCostExpr = "COALESCE(SUM(total_cost * COALESCE(account_rate_multiplier, 1)), 0) as actual_cost"
+	}
+
+	query := fmt.Sprintf(`
+		SELECT
+			CONCAT(
+				COALESCE(NULLIF(TRIM(inbound_endpoint), ''), 'unknown'),
+				' -> ',
+				COALESCE(NULLIF(TRIM(upstream_endpoint), ''), 'unknown')
+			) AS endpoint,
+			COUNT(*) AS requests,
+			COALESCE(SUM(input_tokens + output_tokens + cache_creation_tokens + cache_read_tokens), 0) AS total_tokens,
+			COALESCE(SUM(total_cost), 0) as cost,
+			%s
+		FROM usage_logs
+		WHERE created_at >= $1 AND created_at < $2
+	`, actualCostExpr)
+
+	args := []any{startTime, endTime}
+	if userID > 0 {
+		query += fmt.Sprintf(" AND user_id = $%d", len(args)+1)
+		args = append(args, userID)
+	}
+	if apiKeyID > 0 {
+		query += fmt.Sprintf(" AND api_key_id = $%d", len(args)+1)
+		args = append(args, apiKeyID)
+	}
+	if accountID > 0 {
+		query += fmt.Sprintf(" AND account_id = $%d", len(args)+1)
+		args = append(args, accountID)
+	}
+	if groupID > 0 {
+		query += fmt.Sprintf(" AND group_id = $%d", len(args)+1)
+		args = append(args, groupID)
+	}
+	if model != "" {
+		query += fmt.Sprintf(" AND model = $%d", len(args)+1)
+		args = append(args, model)
+	}
+	query, args = appendRequestTypeOrStreamQueryFilter(query, args, requestType, stream)
+	if billingType != nil {
+		query += fmt.Sprintf(" AND billing_type = $%d", len(args)+1)
+		args = append(args, int16(*billingType))
+	}
+	query += " GROUP BY endpoint ORDER BY requests DESC"
+
+	rows, err := r.sql.QueryContext(ctx, query, args...)
+	if err != nil {
+		return nil, err
+	}
+	defer func() {
+		if closeErr := rows.Close(); closeErr != nil && err == nil {
+			err = closeErr
+			results = nil
+		}
+	}()
+
+	results = make([]EndpointStat, 0)
+	for rows.Next() {
+		var row EndpointStat
+		if err := rows.Scan(&row.Endpoint, &row.Requests, &row.TotalTokens, &row.Cost, &row.ActualCost); err != nil {
+			return nil, err
+		}
+		results = append(results, row)
+	}
+	if err := rows.Err(); err != nil {
+		return nil, err
+	}
+	return results, nil
+}
+
+// GetEndpointStatsWithFilters returns inbound endpoint statistics with optional filters.
+func (r *usageLogRepository) GetEndpointStatsWithFilters(ctx context.Context, startTime, endTime time.Time, userID, apiKeyID, accountID, groupID int64, model string, requestType *int16, stream *bool, billingType *int8) ([]EndpointStat, error) {
+	return r.getEndpointStatsByColumnWithFilters(ctx, "inbound_endpoint", startTime, endTime, userID, apiKeyID, accountID, groupID, model, requestType, stream, billingType)
+}
+
+// GetUpstreamEndpointStatsWithFilters returns upstream endpoint statistics with optional filters.
+func (r *usageLogRepository) GetUpstreamEndpointStatsWithFilters(ctx context.Context, startTime, endTime time.Time, userID, apiKeyID, accountID, groupID int64, model string, requestType *int16, stream *bool, billingType *int8) ([]EndpointStat, error) {
+	return r.getEndpointStatsByColumnWithFilters(ctx, "upstream_endpoint", startTime, endTime, userID, apiKeyID, accountID, groupID, model, requestType, stream, billingType)
+}
+
 // GetAccountUsageStats returns comprehensive usage statistics for an account over a time range
 func (r *usageLogRepository) GetAccountUsageStats(ctx context.Context, accountID int64, startTime, endTime time.Time) (resp *AccountUsageStatsResponse, err error) {
 	daysCount := int(endTime.Sub(startTime).Hours()/24) + 1
@@ -3254,11 +3462,23 @@ func (r *usageLogRepository) GetAccountUsageStats(ctx context.Context, accountID
 	if err != nil {
 		models = []ModelStat{}
 	}
+	endpoints, endpointErr := r.GetEndpointStatsWithFilters(ctx, startTime, endTime, 0, 0, accountID, 0, "", nil, nil, nil)
+	if endpointErr != nil {
+		logger.LegacyPrintf("repository.usage_log", "GetEndpointStatsWithFilters failed in GetAccountUsageStats: %v", endpointErr)
+		endpoints = []EndpointStat{}
+	}
+	upstreamEndpoints, upstreamEndpointErr := r.GetUpstreamEndpointStatsWithFilters(ctx, startTime, endTime, 0, 0, accountID, 0, "", nil, nil, nil)
+	if upstreamEndpointErr != nil {
+		logger.LegacyPrintf("repository.usage_log", "GetUpstreamEndpointStatsWithFilters failed in GetAccountUsageStats: %v", upstreamEndpointErr)
+		upstreamEndpoints = []EndpointStat{}
+	}
 
 	resp = &AccountUsageStatsResponse{
-		History: history,
-		Summary: summary,
-		Models:  models,
+		History:           history,
+		Summary:           summary,
+		Models:            models,
+		Endpoints:         endpoints,
+		UpstreamEndpoints: upstreamEndpoints,
 	}
 	return resp, nil
 }
@@ -3541,6 +3761,8 @@ func scanUsageLog(scanner interface{ Scan(...any) error }) (*service.UsageLog, e
 		mediaType             sql.NullString
 		serviceTier           sql.NullString
 		reasoningEffort       sql.NullString
+		inboundEndpoint       sql.NullString
+		upstreamEndpoint      sql.NullString
 		cacheTTLOverridden    bool
 		createdAt             time.Time
 	)
@@ -3581,6 +3803,8 @@ func scanUsageLog(scanner interface{ Scan(...any) error }) (*service.UsageLog, e
 		&mediaType,
 		&serviceTier,
 		&reasoningEffort,
+		&inboundEndpoint,
+		&upstreamEndpoint,
 		&cacheTTLOverridden,
 		&createdAt,
 	); err != nil {
@@ -3656,6 +3880,12 @@ func scanUsageLog(scanner interface{ Scan(...any) error }) (*service.UsageLog, e
 	if reasoningEffort.Valid {
 		log.ReasoningEffort = &reasoningEffort.String
 	}
+	if inboundEndpoint.Valid {
+		log.InboundEndpoint = &inboundEndpoint.String
+	}
+	if upstreamEndpoint.Valid {
+		log.UpstreamEndpoint = &upstreamEndpoint.String
+	}
 
 	return log, nil
 }

backend/internal/repository/usage_log_repo_request_type_test.go

@@ -73,6 +73,8 @@ func TestUsageLogRepositoryCreateSyncRequestTypeAndLegacyFields(t *testing.T) {
 			sqlmock.AnyArg(), // media_type
 			sqlmock.AnyArg(), // service_tier
 			sqlmock.AnyArg(), // reasoning_effort
+			sqlmock.AnyArg(), // inbound_endpoint
+			sqlmock.AnyArg(), // upstream_endpoint
 			log.CacheTTLOverridden,
 			createdAt,
 		).
@@ -141,6 +143,8 @@ func TestUsageLogRepositoryCreate_PersistsServiceTier(t *testing.T) {
 			sqlmock.AnyArg(),
 			serviceTier,
 			sqlmock.AnyArg(),
+			sqlmock.AnyArg(),
+			sqlmock.AnyArg(),
 			log.CacheTTLOverridden,
 			createdAt,
 		).
@@ -376,6 +380,8 @@ func TestScanUsageLogRequestTypeAndLegacyFallback(t *testing.T) {
 			sql.NullString{},
 			sql.NullString{Valid: true, String: "priority"},
 			sql.NullString{},
+			sql.NullString{},
+			sql.NullString{},
 			false,
 			now,
 		}})
@@ -415,6 +421,8 @@ func TestScanUsageLogRequestTypeAndLegacyFallback(t *testing.T) {
 			sql.NullString{},
 			sql.NullString{Valid: true, String: "flex"},
 			sql.NullString{},
+			sql.NullString{},
+			sql.NullString{},
 			false,
 			now,
 		}})
@@ -454,6 +462,8 @@ func TestScanUsageLogRequestTypeAndLegacyFallback(t *testing.T) {
 			sql.NullString{},
 			sql.NullString{Valid: true, String: "priority"},
 			sql.NullString{},
+			sql.NullString{},
+			sql.NullString{},
 			false,
 			now,
 		}})

backend/internal/server/api_contract_test.go

@@ -493,6 +493,7 @@ func TestAPIContracts(t *testing.T) {
 					"registration_email_suffix_whitelist": [],
 					"promo_code_enabled": true,
 					"password_reset_enabled": false,
+					"frontend_url": "",
 					"totp_enabled": false,
 					"totp_encryption_key_configured": false,
 					"smtp_host": "smtp.example.com",
@@ -1624,6 +1625,14 @@ func (r *stubUsageLogRepo) GetModelStatsWithFilters(ctx context.Context, startTi
 	return nil, errors.New("not implemented")
 }
 
+func (r *stubUsageLogRepo) GetEndpointStatsWithFilters(ctx context.Context, startTime, endTime time.Time, userID, apiKeyID, accountID, groupID int64, model string, requestType *int16, stream *bool, billingType *int8) ([]usagestats.EndpointStat, error) {
+	return nil, errors.New("not implemented")
+}
+
+func (r *stubUsageLogRepo) GetUpstreamEndpointStatsWithFilters(ctx context.Context, startTime, endTime time.Time, userID, apiKeyID, accountID, groupID int64, model string, requestType *int16, stream *bool, billingType *int8) ([]usagestats.EndpointStat, error) {
+	return nil, errors.New("not implemented")
+}
+
 func (r *stubUsageLogRepo) GetGroupStatsWithFilters(ctx context.Context, startTime, endTime time.Time, userID, apiKeyID, accountID, groupID int64, requestType *int16, stream *bool, billingType *int8) ([]usagestats.GroupStat, error) {
 	return nil, errors.New("not implemented")
 }

backend/internal/service/account_usage_service.go

@@ -45,6 +45,8 @@ type UsageLogRepository interface {
 	GetDashboardStats(ctx context.Context) (*usagestats.DashboardStats, error)
 	GetUsageTrendWithFilters(ctx context.Context, startTime, endTime time.Time, granularity string, userID, apiKeyID, accountID, groupID int64, model string, requestType *int16, stream *bool, billingType *int8) ([]usagestats.TrendDataPoint, error)
 	GetModelStatsWithFilters(ctx context.Context, startTime, endTime time.Time, userID, apiKeyID, accountID, groupID int64, requestType *int16, stream *bool, billingType *int8) ([]usagestats.ModelStat, error)
+	GetEndpointStatsWithFilters(ctx context.Context, startTime, endTime time.Time, userID, apiKeyID, accountID, groupID int64, model string, requestType *int16, stream *bool, billingType *int8) ([]usagestats.EndpointStat, error)
+	GetUpstreamEndpointStatsWithFilters(ctx context.Context, startTime, endTime time.Time, userID, apiKeyID, accountID, groupID int64, model string, requestType *int16, stream *bool, billingType *int8) ([]usagestats.EndpointStat, error)
 	GetGroupStatsWithFilters(ctx context.Context, startTime, endTime time.Time, userID, apiKeyID, accountID, groupID int64, requestType *int16, stream *bool, billingType *int8) ([]usagestats.GroupStat, error)
 	GetAPIKeyUsageTrend(ctx context.Context, startTime, endTime time.Time, granularity string, limit int) ([]usagestats.APIKeyUsageTrendPoint, error)
 	GetUserUsageTrend(ctx context.Context, startTime, endTime time.Time, granularity string, limit int) ([]usagestats.UserUsageTrendPoint, error)

backend/internal/service/admin_service.go

@@ -832,7 +832,7 @@ func (s *adminServiceImpl) CreateGroup(ctx context.Context, input *CreateGroupIn
 		subscriptionType = SubscriptionTypeStandard
 	}
 
-	// 限额字段：0 和 nil 都表示"无限制"
+	// 限额字段：nil/负数 表示"无限制"，0 表示"不允许用量"，正数表示具体限额
 	dailyLimit := normalizeLimit(input.DailyLimitUSD)
 	weeklyLimit := normalizeLimit(input.WeeklyLimitUSD)
 	monthlyLimit := normalizeLimit(input.MonthlyLimitUSD)
@@ -944,9 +944,9 @@ func (s *adminServiceImpl) CreateGroup(ctx context.Context, input *CreateGroupIn
 	return group, nil
 }
 
-// normalizeLimit 将 0 或负数转换为 nil（表示无限制）
+// normalizeLimit 将负数转换为 nil（表示无限制），0 保留（表示限额为零）
 func normalizeLimit(limit *float64) *float64 {
-	if limit == nil || *limit <= 0 {
+	if limit == nil || *limit < 0 {
 		return nil
 	}
 	return limit
@@ -1058,16 +1058,11 @@ func (s *adminServiceImpl) UpdateGroup(ctx context.Context, id int64, input *Upd
 	if input.SubscriptionType != "" {
 		group.SubscriptionType = input.SubscriptionType
 	}
-	// 限额字段：0 和 nil 都表示"无限制"，正数表示具体限额
-	if input.DailyLimitUSD != nil {
-		group.DailyLimitUSD = normalizeLimit(input.DailyLimitUSD)
-	}
-	if input.WeeklyLimitUSD != nil {
-		group.WeeklyLimitUSD = normalizeLimit(input.WeeklyLimitUSD)
-	}
-	if input.MonthlyLimitUSD != nil {
-		group.MonthlyLimitUSD = normalizeLimit(input.MonthlyLimitUSD)
-	}
+	// 限额字段：nil/负数 表示"无限制"，0 表示"不允许用量"，正数表示具体限额
+	// 前端始终发送这三个字段，无需 nil 守卫
+	group.DailyLimitUSD = normalizeLimit(input.DailyLimitUSD)
+	group.WeeklyLimitUSD = normalizeLimit(input.WeeklyLimitUSD)
+	group.MonthlyLimitUSD = normalizeLimit(input.MonthlyLimitUSD)
 	// 图片生成计费配置：负数表示清除（使用默认价格）
 	if input.ImagePrice1K != nil {
 		group.ImagePrice1K = normalizePrice(input.ImagePrice1K)

backend/internal/service/api_key.go

@@ -22,8 +22,9 @@ const (
 )
 
 // IsWindowExpired returns true if the window starting at windowStart has exceeded the given duration.
+// A nil windowStart is treated as expired — no initialized window means any accumulated usage is stale.
 func IsWindowExpired(windowStart *time.Time, duration time.Duration) bool {
-	return windowStart != nil && time.Since(*windowStart) >= duration
+	return windowStart == nil || time.Since(*windowStart) >= duration
 }
 
 type APIKey struct {

backend/internal/service/api_key_rate_limit_test.go

@@ -15,10 +15,10 @@ func TestIsWindowExpired(t *testing.T) {
 		want     bool
 	}{
 		{
-			name:     "nil window start",
+			name:     "nil window start (treated as expired)",
 			start:    nil,
 			duration: RateLimitWindow5h,
-			want:     false,
+			want:     true,
 		},
 		{
 			name:     "active window (started 1h ago, 5h window)",
@@ -113,7 +113,7 @@ func TestAPIKey_EffectiveUsage(t *testing.T) {
 			want7d: 0,
 		},
 		{
-			name: "nil window starts return raw usage",
+			name: "nil window starts return 0 (stale usage reset)",
 			key: APIKey{
 				Usage5h:       5.0,
 				Usage1d:       10.0,
@@ -122,9 +122,9 @@ func TestAPIKey_EffectiveUsage(t *testing.T) {
 				Window1dStart: nil,
 				Window7dStart: nil,
 			},
-			want5h: 5.0,
-			want1d: 10.0,
-			want7d: 50.0,
+			want5h: 0,
+			want1d: 0,
+			want7d: 0,
 		},
 		{
 			name: "mixed: 5h expired, 1d active, 7d nil",
@@ -138,7 +138,7 @@ func TestAPIKey_EffectiveUsage(t *testing.T) {
 			},
 			want5h: 0,
 			want1d: 10.0,
-			want7d: 50.0,
+			want7d: 0,
 		},
 		{
 			name: "zero usage with active windows",
@@ -210,7 +210,7 @@ func TestAPIKeyRateLimitData_EffectiveUsage(t *testing.T) {
 			want7d: 0,
 		},
 		{
-			name: "nil window starts return raw usage",
+			name: "nil window starts return 0 (stale usage reset)",
 			data: APIKeyRateLimitData{
 				Usage5h:       3.0,
 				Usage1d:       8.0,
@@ -219,9 +219,9 @@ func TestAPIKeyRateLimitData_EffectiveUsage(t *testing.T) {
 				Window1dStart: nil,
 				Window7dStart: nil,
 			},
-			want5h: 3.0,
-			want1d: 8.0,
-			want7d: 40.0,
+			want5h: 0,
+			want1d: 0,
+			want7d: 0,
 		},
 	}
 

backend/internal/service/domain_constants.go

@@ -80,6 +80,7 @@ const (
 	SettingKeyRegistrationEmailSuffixWhitelist = "registration_email_suffix_whitelist" // 注册邮箱后缀白名单（JSON 数组）
 	SettingKeyPromoCodeEnabled                 = "promo_code_enabled"                  // 是否启用优惠码功能
 	SettingKeyPasswordResetEnabled             = "password_reset_enabled"              // 是否启用忘记密码功能（需要先开启邮件验证）
+	SettingKeyFrontendURL                      = "frontend_url"                        // 前端基础URL，用于生成邮件中的重置密码链接
 	SettingKeyInvitationCodeEnabled            = "invitation_code_enabled"             // 是否启用邀请码注册
 
 	// 邮件服务设置

backend/internal/service/gateway_multiplatform_test.go

@@ -440,7 +440,7 @@ func TestGatewayService_SelectAccountForModelWithPlatform_NoAvailableAccounts(t
 	acc, err := svc.selectAccountForModelWithPlatform(ctx, nil, "", "claude-3-5-sonnet-20241022", nil, PlatformAnthropic)
 	require.Error(t, err)
 	require.Nil(t, acc)
-	require.Contains(t, err.Error(), "no available accounts")
+	require.ErrorIs(t, err, ErrNoAvailableAccounts)
 }
 
 // TestGatewayService_SelectAccountForModelWithPlatform_AllExcluded 测试所有账户被排除
@@ -1073,7 +1073,7 @@ func TestGatewayService_SelectAccountForModelWithPlatform_NoAccounts(t *testing.
 	acc, err := svc.selectAccountForModelWithPlatform(ctx, nil, "", "", nil, PlatformAnthropic)
 	require.Error(t, err)
 	require.Nil(t, acc)
-	require.Contains(t, err.Error(), "no available accounts")
+	require.ErrorIs(t, err, ErrNoAvailableAccounts)
 }
 
 func TestGatewayService_isModelSupportedByAccount(t *testing.T) {
@@ -1734,7 +1734,7 @@ func TestGatewayService_selectAccountWithMixedScheduling(t *testing.T) {
 		acc, err := svc.selectAccountWithMixedScheduling(ctx, nil, "", "claude-3-5-sonnet-20241022", nil, PlatformAnthropic)
 		require.Error(t, err)
 		require.Nil(t, acc)
-		require.Contains(t, err.Error(), "no available accounts")
+		require.ErrorIs(t, err, ErrNoAvailableAccounts)
 	})
 
 	t.Run("混合调度-不支持模型返回错误", func(t *testing.T) {
@@ -2290,7 +2290,7 @@ func TestGatewayService_SelectAccountWithLoadAwareness(t *testing.T) {
 		result, err := svc.SelectAccountWithLoadAwareness(ctx, nil, "", "claude-3-5-sonnet-20241022", nil, "")
 		require.Error(t, err)
 		require.Nil(t, result)
-		require.Contains(t, err.Error(), "no available accounts")
+		require.ErrorIs(t, err, ErrNoAvailableAccounts)
 	})
 
 	t.Run("过滤不可调度账号-限流账号被跳过", func(t *testing.T) {

backend/internal/service/gateway_record_usage_test.go

@@ -369,3 +369,54 @@ func TestGatewayServiceRecordUsage_BillingErrorSkipsUsageLogWrite(t *testing.T)
 	require.Equal(t, 1, billingRepo.calls)
 	require.Equal(t, 0, usageRepo.calls)
 }
+
+func TestGatewayServiceRecordUsage_ReasoningEffortPersisted(t *testing.T) {
+	usageRepo := &openAIRecordUsageBestEffortLogRepoStub{}
+	svc := newGatewayRecordUsageServiceForTest(usageRepo, &openAIRecordUsageUserRepoStub{}, &openAIRecordUsageSubRepoStub{})
+
+	effort := "max"
+	err := svc.RecordUsage(context.Background(), &RecordUsageInput{
+		Result: &ForwardResult{
+			RequestID: "effort_test",
+			Usage: ClaudeUsage{
+				InputTokens:  10,
+				OutputTokens: 5,
+			},
+			Model:           "claude-opus-4-6",
+			Duration:        time.Second,
+			ReasoningEffort: &effort,
+		},
+		APIKey:  &APIKey{ID: 1},
+		User:    &User{ID: 1},
+		Account: &Account{ID: 1},
+	})
+
+	require.NoError(t, err)
+	require.NotNil(t, usageRepo.lastLog)
+	require.NotNil(t, usageRepo.lastLog.ReasoningEffort)
+	require.Equal(t, "max", *usageRepo.lastLog.ReasoningEffort)
+}
+
+func TestGatewayServiceRecordUsage_ReasoningEffortNil(t *testing.T) {
+	usageRepo := &openAIRecordUsageBestEffortLogRepoStub{}
+	svc := newGatewayRecordUsageServiceForTest(usageRepo, &openAIRecordUsageUserRepoStub{}, &openAIRecordUsageSubRepoStub{})
+
+	err := svc.RecordUsage(context.Background(), &RecordUsageInput{
+		Result: &ForwardResult{
+			RequestID: "no_effort_test",
+			Usage: ClaudeUsage{
+				InputTokens:  10,
+				OutputTokens: 5,
+			},
+			Model:    "claude-sonnet-4",
+			Duration: time.Second,
+		},
+		APIKey:  &APIKey{ID: 1},
+		User:    &User{ID: 1},
+		Account: &Account{ID: 1},
+	})
+
+	require.NoError(t, err)
+	require.NotNil(t, usageRepo.lastLog)
+	require.Nil(t, usageRepo.lastLog.ReasoningEffort)
+}

backend/internal/service/gateway_request.go

@@ -60,6 +60,7 @@ type ParsedRequest struct {
 	Messages        []any           // messages 数组
 	HasSystem       bool            // 是否包含 system 字段（包含 null 也视为显式传入）
 	ThinkingEnabled bool            // 是否开启 thinking（部分平台会影响最终模型名）
+	OutputEffort    string          // output_config.effort（Claude API 的推理强度控制）
 	MaxTokens       int             // max_tokens 值（用于探测请求拦截）
 	SessionContext  *SessionContext // 可选：请求上下文区分因子（nil 时行为不变）
 
@@ -116,6 +117,9 @@ func ParseGatewayRequest(body []byte, protocol string) (*ParsedRequest, error) {
 		parsed.ThinkingEnabled = true
 	}
 
+	// output_config.effort: Claude API 的推理强度控制参数
+	parsed.OutputEffort = strings.TrimSpace(gjson.Get(jsonStr, "output_config.effort").String())
+
 	// max_tokens: 仅接受整数值
 	maxTokensResult := gjson.Get(jsonStr, "max_tokens")
 	if maxTokensResult.Exists() && maxTokensResult.Type == gjson.Number {
@@ -747,6 +751,21 @@ func filterThinkingBlocksInternal(body []byte, _ bool) []byte {
 	return newBody
 }
 
+// NormalizeClaudeOutputEffort normalizes Claude's output_config.effort value.
+// Returns nil for empty or unrecognized values.
+func NormalizeClaudeOutputEffort(raw string) *string {
+	value := strings.ToLower(strings.TrimSpace(raw))
+	if value == "" {
+		return nil
+	}
+	switch value {
+	case "low", "medium", "high", "max":
+		return &value
+	default:
+		return nil
+	}
+}
+
 // =========================
 // Thinking Budget Rectifier
 // =========================

backend/internal/service/gateway_request_test.go

@@ -972,6 +972,76 @@ func BenchmarkParseGatewayRequest_Old_Large(b *testing.B) {
 	}
 }
 
+func TestParseGatewayRequest_OutputEffort(t *testing.T) {
+	tests := []struct {
+		name       string
+		body       string
+		wantEffort string
+	}{
+		{
+			name:       "output_config.effort present",
+			body:       `{"model":"claude-opus-4-6","output_config":{"effort":"medium"},"messages":[]}`,
+			wantEffort: "medium",
+		},
+		{
+			name:       "output_config.effort max",
+			body:       `{"model":"claude-opus-4-6","output_config":{"effort":"max"},"messages":[]}`,
+			wantEffort: "max",
+		},
+		{
+			name:       "output_config without effort",
+			body:       `{"model":"claude-opus-4-6","output_config":{},"messages":[]}`,
+			wantEffort: "",
+		},
+		{
+			name:       "no output_config",
+			body:       `{"model":"claude-opus-4-6","messages":[]}`,
+			wantEffort: "",
+		},
+		{
+			name:       "effort with whitespace trimmed",
+			body:       `{"model":"claude-opus-4-6","output_config":{"effort":" high "},"messages":[]}`,
+			wantEffort: "high",
+		},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			parsed, err := ParseGatewayRequest([]byte(tt.body), "")
+			require.NoError(t, err)
+			require.Equal(t, tt.wantEffort, parsed.OutputEffort)
+		})
+	}
+}
+
+func TestNormalizeClaudeOutputEffort(t *testing.T) {
+	tests := []struct {
+		input string
+		want  *string
+	}{
+		{"low", strPtr("low")},
+		{"medium", strPtr("medium")},
+		{"high", strPtr("high")},
+		{"max", strPtr("max")},
+		{"LOW", strPtr("low")},
+		{"Max", strPtr("max")},
+		{" medium ", strPtr("medium")},
+		{"", nil},
+		{"unknown", nil},
+		{"xhigh", nil},
+	}
+	for _, tt := range tests {
+		t.Run(tt.input, func(t *testing.T) {
+			got := NormalizeClaudeOutputEffort(tt.input)
+			if tt.want == nil {
+				require.Nil(t, got)
+			} else {
+				require.NotNil(t, got)
+				require.Equal(t, *tt.want, *got)
+			}
+		})
+	}
+}
+
 func BenchmarkParseGatewayRequest_New_Large(b *testing.B) {
 	data := buildLargeJSON()
 	b.SetBytes(int64(len(data)))

backend/internal/service/gateway_service.go

@@ -346,6 +346,9 @@ var systemBlockFilterPrefixes = []string{
 	"x-anthropic-billing-header",
 }
 
+// ErrNoAvailableAccounts 表示没有可用的账号
+var ErrNoAvailableAccounts = errors.New("no available accounts")
+
 // ErrClaudeCodeOnly 表示分组仅允许 Claude Code 客户端访问
 var ErrClaudeCodeOnly = errors.New("this group only allows Claude Code clients")
 
@@ -492,6 +495,7 @@ type ForwardResult struct {
 	Duration         time.Duration
 	FirstTokenMs     *int // 首字时间（流式请求）
 	ClientDisconnect bool // 客户端是否在流式传输过程中断开
+	ReasoningEffort  *string
 
 	// 图片生成计费字段（图片生成模型使用）
 	ImageCount int    // 生成的图片数量
@@ -1204,7 +1208,7 @@ func (s *GatewayService) SelectAccountWithLoadAwareness(ctx context.Context, gro
 		return nil, err
 	}
 	if len(accounts) == 0 {
-		return nil, errors.New("no available accounts")
+		return nil, ErrNoAvailableAccounts
 	}
 	ctx = s.withWindowCostPrefetch(ctx, accounts)
 	ctx = s.withRPMPrefetch(ctx, accounts)
@@ -1552,7 +1556,7 @@ func (s *GatewayService) SelectAccountWithLoadAwareness(ctx context.Context, gro
 	}
 
 	if len(candidates) == 0 {
-		return nil, errors.New("no available accounts")
+		return nil, ErrNoAvailableAccounts
 	}
 
 	accountLoads := make([]AccountWithConcurrency, 0, len(candidates))
@@ -1641,7 +1645,7 @@ func (s *GatewayService) SelectAccountWithLoadAwareness(ctx context.Context, gro
 			},
 		}, nil
 	}
-	return nil, errors.New("no available accounts")
+	return nil, ErrNoAvailableAccounts
 }
 
 func (s *GatewayService) tryAcquireByLegacyOrder(ctx context.Context, candidates []*Account, groupID *int64, sessionHash string, preferOAuth bool) (*AccountSelectionResult, bool) {
@@ -2851,9 +2855,9 @@ func (s *GatewayService) selectAccountForModelWithPlatform(ctx context.Context,
 	if selected == nil {
 		stats := s.logDetailedSelectionFailure(ctx, groupID, sessionHash, requestedModel, platform, accounts, excludedIDs, false)
 		if requestedModel != "" {
-			return nil, fmt.Errorf("no available accounts supporting model: %s (%s)", requestedModel, summarizeSelectionFailureStats(stats))
+			return nil, fmt.Errorf("%w supporting model: %s (%s)", ErrNoAvailableAccounts, requestedModel, summarizeSelectionFailureStats(stats))
 		}
-		return nil, errors.New("no available accounts")
+		return nil, ErrNoAvailableAccounts
 	}
 
 	// 4. 建立粘性绑定
@@ -3089,9 +3093,9 @@ func (s *GatewayService) selectAccountWithMixedScheduling(ctx context.Context, g
 	if selected == nil {
 		stats := s.logDetailedSelectionFailure(ctx, groupID, sessionHash, requestedModel, nativePlatform, accounts, excludedIDs, true)
 		if requestedModel != "" {
-			return nil, fmt.Errorf("no available accounts supporting model: %s (%s)", requestedModel, summarizeSelectionFailureStats(stats))
+			return nil, fmt.Errorf("%w supporting model: %s (%s)", ErrNoAvailableAccounts, requestedModel, summarizeSelectionFailureStats(stats))
 		}
-		return nil, errors.New("no available accounts")
+		return nil, ErrNoAvailableAccounts
 	}
 
 	// 4. 建立粘性绑定
@@ -7523,6 +7527,7 @@ func (s *GatewayService) RecordUsage(ctx context.Context, input *RecordUsageInpu
 		AccountID:             account.ID,
 		RequestID:             requestID,
 		Model:                 result.Model,
+		ReasoningEffort:       result.ReasoningEffort,
 		InputTokens:           result.Usage.InputTokens,
 		OutputTokens:          result.Usage.OutputTokens,
 		CacheCreationTokens:   result.Usage.CacheCreationInputTokens,
@@ -7699,6 +7704,7 @@ func (s *GatewayService) RecordUsageWithLongContext(ctx context.Context, input *
 		AccountID:             account.ID,
 		RequestID:             requestID,
 		Model:                 result.Model,
+		ReasoningEffort:       result.ReasoningEffort,
 		InputTokens:           result.Usage.InputTokens,
 		OutputTokens:          result.Usage.OutputTokens,
 		CacheCreationTokens:   result.Usage.CacheCreationInputTokens,

backend/internal/service/gemini_messages_compat_service.go

@@ -3235,7 +3235,7 @@ func cleanToolSchema(schema any) any {
 		for key, value := range v {
 			// 跳过不支持的字段
 			if key == "$schema" || key == "$id" || key == "$ref" ||
-				key == "additionalProperties" || key == "minLength" ||
+				key == "additionalProperties" || key == "patternProperties" || key == "minLength" ||
 				key == "maxLength" || key == "minItems" || key == "maxItems" {
 				continue
 			}

backend/internal/service/openai_account_scheduler.go

@@ -725,7 +725,7 @@ func (s *defaultOpenAIAccountScheduler) selectByLoadBalance(
 		}, len(candidates), topK, loadSkew, nil
 	}
 
-	return nil, len(candidates), topK, loadSkew, errors.New("no available accounts")
+	return nil, len(candidates), topK, loadSkew, ErrNoAvailableAccounts
 }
 
 func (s *defaultOpenAIAccountScheduler) isAccountTransportCompatible(account *Account, requiredTransport OpenAIUpstreamTransport) bool {

backend/internal/service/openai_gateway_record_usage_test.go

@@ -226,6 +226,41 @@ func TestOpenAIGatewayServiceRecordUsage_UsesUserSpecificGroupRate(t *testing.T)
 	require.Equal(t, 1, userRepo.deductCalls)
 }
 
+func TestOpenAIGatewayServiceRecordUsage_IncludesEndpointMetadata(t *testing.T) {
+	usageRepo := &openAIRecordUsageLogRepoStub{inserted: true}
+	userRepo := &openAIRecordUsageUserRepoStub{}
+	subRepo := &openAIRecordUsageSubRepoStub{}
+	rateRepo := &openAIUserGroupRateRepoStub{}
+	svc := newOpenAIRecordUsageServiceForTest(usageRepo, userRepo, subRepo, rateRepo)
+
+	err := svc.RecordUsage(context.Background(), &OpenAIRecordUsageInput{
+		Result: &OpenAIForwardResult{
+			RequestID: "resp_endpoint_metadata",
+			Usage: OpenAIUsage{
+				InputTokens:  8,
+				OutputTokens: 2,
+			},
+			Model:    "gpt-5.1",
+			Duration: time.Second,
+		},
+		APIKey: &APIKey{
+			ID:    1002,
+			Group: &Group{RateMultiplier: 1},
+		},
+		User:             &User{ID: 2002},
+		Account:          &Account{ID: 3002},
+		InboundEndpoint:  " /v1/chat/completions ",
+		UpstreamEndpoint: " /v1/responses ",
+	})
+
+	require.NoError(t, err)
+	require.NotNil(t, usageRepo.lastLog)
+	require.NotNil(t, usageRepo.lastLog.InboundEndpoint)
+	require.Equal(t, "/v1/chat/completions", *usageRepo.lastLog.InboundEndpoint)
+	require.NotNil(t, usageRepo.lastLog.UpstreamEndpoint)
+	require.Equal(t, "/v1/responses", *usageRepo.lastLog.UpstreamEndpoint)
+}
+
 func TestOpenAIGatewayServiceRecordUsage_FallsBackToGroupDefaultRateOnResolverError(t *testing.T) {
 	groupID := int64(12)
 	groupRate := 1.6

backend/internal/service/openai_gateway_service.go

@@ -1312,7 +1312,7 @@ func (s *OpenAIGatewayService) SelectAccountWithLoadAwareness(ctx context.Contex
 		return nil, err
 	}
 	if len(accounts) == 0 {
-		return nil, errors.New("no available accounts")
+		return nil, ErrNoAvailableAccounts
 	}
 
 	isExcluded := func(accountID int64) bool {
@@ -1382,7 +1382,7 @@ func (s *OpenAIGatewayService) SelectAccountWithLoadAwareness(ctx context.Contex
 	}
 
 	if len(candidates) == 0 {
-		return nil, errors.New("no available accounts")
+		return nil, ErrNoAvailableAccounts
 	}
 
 	accountLoads := make([]AccountWithConcurrency, 0, len(candidates))
@@ -1489,7 +1489,7 @@ func (s *OpenAIGatewayService) SelectAccountWithLoadAwareness(ctx context.Contex
 		}, nil
 	}
 
-	return nil, errors.New("no available accounts")
+	return nil, ErrNoAvailableAccounts
 }
 
 func (s *OpenAIGatewayService) listSchedulableAccounts(ctx context.Context, groupID *int64) ([]Account, error) {
@@ -4028,6 +4028,8 @@ type OpenAIRecordUsageInput struct {
 	User               *User
 	Account            *Account
 	Subscription       *UserSubscription
+	InboundEndpoint    string
+	UpstreamEndpoint   string
 	UserAgent          string // 请求的 User-Agent
 	IPAddress          string // 请求的客户端 IP 地址
 	RequestPayloadHash string
@@ -4106,6 +4108,8 @@ func (s *OpenAIGatewayService) RecordUsage(ctx context.Context, input *OpenAIRec
 		Model:                 billingModel,
 		ServiceTier:           result.ServiceTier,
 		ReasoningEffort:       result.ReasoningEffort,
+		InboundEndpoint:       optionalTrimmedStringPtr(input.InboundEndpoint),
+		UpstreamEndpoint:      optionalTrimmedStringPtr(input.UpstreamEndpoint),
 		InputTokens:           actualInputTokens,
 		OutputTokens:          result.Usage.OutputTokens,
 		CacheCreationTokens:   result.Usage.CacheCreationInputTokens,
@@ -4125,7 +4129,6 @@ func (s *OpenAIGatewayService) RecordUsage(ctx context.Context, input *OpenAIRec
 		FirstTokenMs:          result.FirstTokenMs,
 		CreatedAt:             time.Now(),
 	}
-
 	// 添加 UserAgent
 	if input.UserAgent != "" {
 		usageLog.UserAgent = &input.UserAgent
@@ -4668,3 +4671,11 @@ func normalizeOpenAIReasoningEffort(raw string) string {
 		return ""
 	}
 }
+
+func optionalTrimmedStringPtr(raw string) *string {
+	trimmed := strings.TrimSpace(raw)
+	if trimmed == "" {
+		return nil
+	}
+	return &trimmed
+}

backend/internal/service/ops_retry.go

@@ -467,7 +467,7 @@ func (s *OpsService) executeClientRetry(ctx context.Context, reqType opsRetryReq
 			return &opsRetryExecution{status: opsRetryStatusFailed, errorMessage: selErr.Error()}
 		}
 		if selection == nil || selection.Account == nil {
-			return &opsRetryExecution{status: opsRetryStatusFailed, errorMessage: "no available accounts"}
+			return &opsRetryExecution{status: opsRetryStatusFailed, errorMessage: ErrNoAvailableAccounts.Error()}
 		}
 
 		account := selection.Account

backend/internal/service/ops_settings.go

@@ -368,13 +368,14 @@ func defaultOpsAdvancedSettings() *OpsAdvancedSettings {
 		Aggregation: OpsAggregationSettings{
 			AggregationEnabled: false,
 		},
-		IgnoreCountTokensErrors:   true,  // count_tokens 404 是预期行为，默认忽略
-		IgnoreContextCanceled:     true,  // Default to true - client disconnects are not errors
-		IgnoreNoAvailableAccounts: false, // Default to false - this is a real routing issue
-		DisplayOpenAITokenStats:   false,
-		DisplayAlertEvents:        true,
-		AutoRefreshEnabled:        false,
-		AutoRefreshIntervalSec:    30,
+		IgnoreCountTokensErrors:         true,  // count_tokens 404 是预期行为，默认忽略
+		IgnoreContextCanceled:           true,  // Default to true - client disconnects are not errors
+		IgnoreNoAvailableAccounts:       false, // Default to false - this is a real routing issue
+		IgnoreInsufficientBalanceErrors: false, // 默认不忽略，余额不足可能需要关注
+		DisplayOpenAITokenStats:         false,
+		DisplayAlertEvents:              true,
+		AutoRefreshEnabled:              false,
+		AutoRefreshIntervalSec:          30,
 	}
 }
 

backend/internal/service/ops_settings_models.go

@@ -92,16 +92,17 @@ type OpsAlertRuntimeSettings struct {
 
 // OpsAdvancedSettings stores advanced ops configuration (data retention, aggregation).
 type OpsAdvancedSettings struct {
-	DataRetention             OpsDataRetentionSettings `json:"data_retention"`
-	Aggregation               OpsAggregationSettings   `json:"aggregation"`
-	IgnoreCountTokensErrors   bool                     `json:"ignore_count_tokens_errors"`
-	IgnoreContextCanceled     bool                     `json:"ignore_context_canceled"`
-	IgnoreNoAvailableAccounts bool                     `json:"ignore_no_available_accounts"`
-	IgnoreInvalidApiKeyErrors bool                     `json:"ignore_invalid_api_key_errors"`
-	DisplayOpenAITokenStats   bool                     `json:"display_openai_token_stats"`
-	DisplayAlertEvents        bool                     `json:"display_alert_events"`
-	AutoRefreshEnabled        bool                     `json:"auto_refresh_enabled"`
-	AutoRefreshIntervalSec    int                      `json:"auto_refresh_interval_seconds"`
+	DataRetention                   OpsDataRetentionSettings `json:"data_retention"`
+	Aggregation                     OpsAggregationSettings   `json:"aggregation"`
+	IgnoreCountTokensErrors         bool                     `json:"ignore_count_tokens_errors"`
+	IgnoreContextCanceled           bool                     `json:"ignore_context_canceled"`
+	IgnoreNoAvailableAccounts       bool                     `json:"ignore_no_available_accounts"`
+	IgnoreInvalidApiKeyErrors       bool                     `json:"ignore_invalid_api_key_errors"`
+	IgnoreInsufficientBalanceErrors bool                     `json:"ignore_insufficient_balance_errors"`
+	DisplayOpenAITokenStats         bool                     `json:"display_openai_token_stats"`
+	DisplayAlertEvents              bool                     `json:"display_alert_events"`
+	AutoRefreshEnabled              bool                     `json:"auto_refresh_enabled"`
+	AutoRefreshIntervalSec          int                      `json:"auto_refresh_interval_seconds"`
 }
 
 type OpsDataRetentionSettings struct {

backend/internal/service/setting_service.go

@@ -116,6 +116,15 @@ func (s *SettingService) GetAllSettings(ctx context.Context) (*SystemSettings, e
 	return s.parseSettings(settings), nil
 }
 
+// GetFrontendURL 获取前端基础URL（数据库优先，fallback 到配置文件）
+func (s *SettingService) GetFrontendURL(ctx context.Context) string {
+	val, err := s.settingRepo.GetValue(ctx, SettingKeyFrontendURL)
+	if err == nil && strings.TrimSpace(val) != "" {
+		return strings.TrimSpace(val)
+	}
+	return s.cfg.Server.FrontendURL
+}
+
 // GetPublicSettings 获取公开设置（无需登录）
 func (s *SettingService) GetPublicSettings(ctx context.Context) (*PublicSettings, error) {
 	keys := []string{
@@ -401,6 +410,7 @@ func (s *SettingService) UpdateSettings(ctx context.Context, settings *SystemSet
 	updates[SettingKeyRegistrationEmailSuffixWhitelist] = string(registrationEmailSuffixWhitelistJSON)
 	updates[SettingKeyPromoCodeEnabled] = strconv.FormatBool(settings.PromoCodeEnabled)
 	updates[SettingKeyPasswordResetEnabled] = strconv.FormatBool(settings.PasswordResetEnabled)
+	updates[SettingKeyFrontendURL] = settings.FrontendURL
 	updates[SettingKeyInvitationCodeEnabled] = strconv.FormatBool(settings.InvitationCodeEnabled)
 	updates[SettingKeyTotpEnabled] = strconv.FormatBool(settings.TotpEnabled)
 
@@ -767,6 +777,7 @@ func (s *SettingService) parseSettings(settings map[string]string) *SystemSettin
 		RegistrationEmailSuffixWhitelist: ParseRegistrationEmailSuffixWhitelist(settings[SettingKeyRegistrationEmailSuffixWhitelist]),
 		PromoCodeEnabled:                 settings[SettingKeyPromoCodeEnabled] != "false", // 默认启用
 		PasswordResetEnabled:             emailVerifyEnabled && settings[SettingKeyPasswordResetEnabled] == "true",
+		FrontendURL:                      settings[SettingKeyFrontendURL],
 		InvitationCodeEnabled:            settings[SettingKeyInvitationCodeEnabled] == "true",
 		TotpEnabled:                      settings[SettingKeyTotpEnabled] == "true",
 		SMTPHost:                         settings[SettingKeySMTPHost],

backend/internal/service/settings_view.go

@@ -6,6 +6,7 @@ type SystemSettings struct {
 	RegistrationEmailSuffixWhitelist []string
 	PromoCodeEnabled                 bool
 	PasswordResetEnabled             bool
+	FrontendURL                      string
 	InvitationCodeEnabled            bool
 	TotpEnabled                      bool // TOTP 双因素认证
 

backend/internal/service/usage_log.go

@@ -100,9 +100,14 @@ type UsageLog struct {
 	Model     string
 	// ServiceTier records the OpenAI service tier used for billing, e.g. "priority" / "flex".
 	ServiceTier *string
-	// ReasoningEffort is the request's reasoning effort level (OpenAI Responses API),
-	// e.g. "low" / "medium" / "high" / "xhigh". Nil means not provided / not applicable.
+	// ReasoningEffort is the request's reasoning effort level.
+	// OpenAI: "low" / "medium" / "high" / "xhigh"; Claude: "low" / "medium" / "high" / "max".
+	// Nil means not provided / not applicable.
 	ReasoningEffort *string
+	// InboundEndpoint is the client-facing API endpoint path, e.g. /v1/chat/completions.
+	InboundEndpoint *string
+	// UpstreamEndpoint is the normalized upstream endpoint path, e.g. /v1/responses.
+	UpstreamEndpoint *string
 
 	GroupID        *int64
 	SubscriptionID *int64

backend/migrations/074_add_usage_log_endpoints.sql

@@ -0,0 +1,5 @@
+-- Add endpoint tracking fields to usage_logs.
+-- inbound_endpoint: client-facing API route (e.g. /v1/chat/completions, /v1/messages, /v1/responses)
+-- upstream_endpoint: normalized upstream route (e.g. /v1/responses)
+ALTER TABLE usage_logs ADD COLUMN IF NOT EXISTS inbound_endpoint VARCHAR(128);
+ALTER TABLE usage_logs ADD COLUMN IF NOT EXISTS upstream_endpoint VARCHAR(128);

frontend/src/api/admin/ops.ts

@@ -841,6 +841,7 @@ export interface OpsAdvancedSettings {
   ignore_context_canceled: boolean
   ignore_no_available_accounts: boolean
   ignore_invalid_api_key_errors: boolean
+  ignore_insufficient_balance_errors: boolean
   display_openai_token_stats: boolean
   display_alert_events: boolean
   auto_refresh_enabled: boolean

frontend/src/api/admin/settings.ts

@@ -21,6 +21,7 @@ export interface SystemSettings {
   registration_email_suffix_whitelist: string[]
   promo_code_enabled: boolean
   password_reset_enabled: boolean
+  frontend_url: string
   invitation_code_enabled: boolean
   totp_enabled: boolean // TOTP 双因素认证
   totp_encryption_key_configured: boolean // TOTP 加密密钥是否已配置
@@ -91,6 +92,7 @@ export interface UpdateSettingsRequest {
   registration_email_suffix_whitelist?: string[]
   promo_code_enabled?: boolean
   password_reset_enabled?: boolean
+  frontend_url?: string
   invitation_code_enabled?: boolean
   totp_enabled?: boolean // TOTP 双因素认证
   default_balance?: number

frontend/src/api/admin/usage.ts

@@ -5,6 +5,7 @@
 
 import { apiClient } from '../client'
 import type { AdminUsageLog, UsageQueryParams, PaginatedResponse, UsageRequestType } from '@/types'
+import type { EndpointStat } from '@/types'
 
 // ==================== Types ====================
 
@@ -18,6 +19,9 @@ export interface AdminUsageStatsResponse {
   total_actual_cost: number
   total_account_cost?: number
   average_duration_ms: number
+  endpoints?: EndpointStat[]
+  upstream_endpoints?: EndpointStat[]
+  endpoint_paths?: EndpointStat[]
 }
 
 export interface SimpleUser {

frontend/src/components/account/AccountStatsModal.vue

@@ -446,6 +446,18 @@
 
         <!-- Model Distribution -->
         <ModelDistributionChart :model-stats="stats.models" :loading="false" />
+
+        <EndpointDistributionChart
+          :endpoint-stats="stats.endpoints || []"
+          :loading="false"
+          :title="t('usage.inboundEndpoint')"
+        />
+
+        <EndpointDistributionChart
+          :endpoint-stats="stats.upstream_endpoints || []"
+          :loading="false"
+          :title="t('usage.upstreamEndpoint')"
+        />
       </template>
 
       <!-- No Data State -->
@@ -489,6 +501,7 @@ import { Line } from 'vue-chartjs'
 import BaseDialog from '@/components/common/BaseDialog.vue'
 import LoadingSpinner from '@/components/common/LoadingSpinner.vue'
 import ModelDistributionChart from '@/components/charts/ModelDistributionChart.vue'
+import EndpointDistributionChart from '@/components/charts/EndpointDistributionChart.vue'
 import Icon from '@/components/icons/Icon.vue'
 import { adminAPI } from '@/api/admin'
 import type { Account, AccountUsageStatsResponse } from '@/types'

frontend/src/components/account/BulkEditAccountModal.vue

@@ -164,27 +164,10 @@
               </p>
             </div>
 
-            <!-- Model Checkbox List -->
-            <div class="mb-3 grid grid-cols-2 gap-2">
-              <label
-                v-for="model in filteredModels"
-                :key="model.value"
-                class="flex cursor-pointer items-center rounded-lg border p-3 transition-all hover:bg-gray-50 dark:border-dark-600 dark:hover:bg-dark-700"
-                :class="
-                  allowedModels.includes(model.value)
-                    ? 'border-primary-500 bg-primary-50 dark:bg-primary-900/20'
-                    : 'border-gray-200'
-                "
-              >
-                <input
-                  v-model="allowedModels"
-                  type="checkbox"
-                  :value="model.value"
-                  class="mr-2 rounded border-gray-300 text-primary-600 focus:ring-primary-500"
-                />
-                <span class="text-sm text-gray-700 dark:text-gray-300">{{ model.label }}</span>
-              </label>
-            </div>
+            <ModelWhitelistSelector
+              v-model="allowedModels"
+              :platforms="selectedPlatforms"
+            />
 
             <p class="text-xs text-gray-500 dark:text-gray-400">
               {{ t('admin.accounts.selectedModels', { count: allowedModels.length }) }}
@@ -832,8 +815,12 @@ import ConfirmDialog from '@/components/common/ConfirmDialog.vue'
 import Select from '@/components/common/Select.vue'
 import ProxySelector from '@/components/common/ProxySelector.vue'
 import GroupSelector from '@/components/common/GroupSelector.vue'
+import ModelWhitelistSelector from '@/components/account/ModelWhitelistSelector.vue'
 import Icon from '@/components/icons/Icon.vue'
-import { buildModelMappingObject as buildModelMappingPayload } from '@/composables/useModelWhitelist'
+import {
+  buildModelMappingObject as buildModelMappingPayload,
+  getPresetMappingsByPlatform
+} from '@/composables/useModelWhitelist'
 
 interface Props {
   show: boolean
@@ -865,26 +852,20 @@ const allAnthropicOAuthOrSetupToken = computed(() => {
   )
 })
 
-const platformModelPrefix: Record<string, string[]> = {
-  anthropic: ['claude-'],
-  antigravity: ['claude-', 'gemini-', 'gpt-oss-', 'tab_'],
-  openai: ['gpt-'],
-  gemini: ['gemini-'],
-  sora: []
-}
-
-const filteredModels = computed(() => {
-  if (props.selectedPlatforms.length === 0) return allModels
-  const prefixes = [...new Set(props.selectedPlatforms.flatMap(p => platformModelPrefix[p] || []))]
-  if (prefixes.length === 0) return allModels
-  return allModels.filter(m => prefixes.some(prefix => m.value.startsWith(prefix)))
-})
-
 const filteredPresets = computed(() => {
-  if (props.selectedPlatforms.length === 0) return presetMappings
-  const prefixes = [...new Set(props.selectedPlatforms.flatMap(p => platformModelPrefix[p] || []))]
-  if (prefixes.length === 0) return presetMappings
-  return presetMappings.filter(m => prefixes.some(prefix => m.from.startsWith(prefix)))
+  if (props.selectedPlatforms.length === 0) return []
+
+  const dedupedPresets = new Map<string, ReturnType<typeof getPresetMappingsByPlatform>[number]>()
+  for (const platform of props.selectedPlatforms) {
+    for (const preset of getPresetMappingsByPlatform(platform)) {
+      const key = `${preset.from}=>${preset.to}`
+      if (!dedupedPresets.has(key)) {
+        dedupedPresets.set(key, preset)
+      }
+    }
+  }
+
+  return Array.from(dedupedPresets.values())
 })
 
 // Model mapping type
@@ -937,204 +918,6 @@ const umqModeOptions = computed(() => [
   { value: 'serialize', label: t('admin.accounts.quotaControl.rpmLimit.umqModeSerialize') },
 ])
 
-// All models list (combined Anthropic + OpenAI + Gemini)
-const allModels = [
-  { value: 'claude-opus-4-6', label: 'Claude Opus 4.6' },
-  { value: 'claude-sonnet-4-6', label: 'Claude Sonnet 4.6' },
-  { value: 'claude-opus-4-5-20251101', label: 'Claude Opus 4.5' },
-  { value: 'claude-sonnet-4-20250514', label: 'Claude Sonnet 4' },
-  { value: 'claude-sonnet-4-5-20250929', label: 'Claude Sonnet 4.5' },
-  { value: 'claude-3-5-haiku-20241022', label: 'Claude 3.5 Haiku' },
-  { value: 'claude-haiku-4-5-20251001', label: 'Claude Haiku 4.5' },
-  { value: 'claude-3-opus-20240229', label: 'Claude 3 Opus' },
-  { value: 'claude-3-5-sonnet-20241022', label: 'Claude 3.5 Sonnet' },
-  { value: 'claude-3-haiku-20240307', label: 'Claude 3 Haiku' },
-  { value: 'gpt-5.3-codex', label: 'GPT-5.3 Codex' },
-  { value: 'gpt-5.3-codex-spark', label: 'GPT-5.3 Codex Spark' },
-  { value: 'gpt-5.4', label: 'GPT-5.4' },
-  { value: 'gpt-5.2-2025-12-11', label: 'GPT-5.2' },
-  { value: 'gpt-5.2-codex', label: 'GPT-5.2 Codex' },
-  { value: 'gpt-5.1-codex-max', label: 'GPT-5.1 Codex Max' },
-  { value: 'gpt-5.1-codex', label: 'GPT-5.1 Codex' },
-  { value: 'gpt-5.1-2025-11-13', label: 'GPT-5.1' },
-  { value: 'gpt-5.1-codex-mini', label: 'GPT-5.1 Codex Mini' },
-  { value: 'gpt-5-2025-08-07', label: 'GPT-5' },
-  { value: 'gemini-3.1-flash-image', label: 'Gemini 3.1 Flash Image' },
-  { value: 'gemini-2.5-flash-image', label: 'Gemini 2.5 Flash Image' },
-  { value: 'gemini-2.0-flash', label: 'Gemini 2.0 Flash' },
-  { value: 'gemini-2.5-flash', label: 'Gemini 2.5 Flash' },
-  { value: 'gemini-2.5-pro', label: 'Gemini 2.5 Pro' },
-  { value: 'gemini-3-pro-image', label: 'Gemini 3 Pro Image (Legacy)' },
-  { value: 'gemini-3-flash-preview', label: 'Gemini 3 Flash Preview' },
-  { value: 'gemini-3-pro-preview', label: 'Gemini 3 Pro Preview' }
-]
-
-// Preset mappings (combined Anthropic + OpenAI + Gemini)
-const presetMappings = [
-  {
-    label: 'Sonnet 4',
-    from: 'claude-sonnet-4-20250514',
-    to: 'claude-sonnet-4-20250514',
-    color: 'bg-blue-100 text-blue-700 hover:bg-blue-200 dark:bg-blue-900/30 dark:text-blue-400'
-  },
-  {
-    label: 'Sonnet 4.5',
-    from: 'claude-sonnet-4-5-20250929',
-    to: 'claude-sonnet-4-5-20250929',
-    color:
-      'bg-indigo-100 text-indigo-700 hover:bg-indigo-200 dark:bg-indigo-900/30 dark:text-indigo-400'
-  },
-  {
-    label: 'Opus 4.5',
-    from: 'claude-opus-4-5-20251101',
-    to: 'claude-opus-4-5-20251101',
-    color:
-      'bg-purple-100 text-purple-700 hover:bg-purple-200 dark:bg-purple-900/30 dark:text-purple-400'
-  },
-  {
-    label: 'Opus 4.6',
-    from: 'claude-opus-4-6',
-    to: 'claude-opus-4-6-thinking',
-    color:
-      'bg-purple-100 text-purple-700 hover:bg-purple-200 dark:bg-purple-900/30 dark:text-purple-400'
-  },
-  {
-    label: 'Opus 4.6-thinking',
-    from: 'claude-opus-4-6-thinking',
-    to: 'claude-opus-4-6-thinking',
-    color:
-      'bg-purple-100 text-purple-700 hover:bg-purple-200 dark:bg-purple-900/30 dark:text-purple-400'
-  },
-  {
-    label: 'Sonnet 4.6',
-    from: 'claude-sonnet-4-6',
-    to: 'claude-sonnet-4-6',
-    color:
-      'bg-purple-100 text-purple-700 hover:bg-purple-200 dark:bg-purple-900/30 dark:text-purple-400'
-  },
-  {
-    label: 'Sonnet4→4.6',
-    from: 'claude-sonnet-4-20250514',
-    to: 'claude-sonnet-4-6',
-    color: 'bg-sky-100 text-sky-700 hover:bg-sky-200 dark:bg-sky-900/30 dark:text-sky-400'
-  },
-  {
-    label: 'Sonnet4.5→4.6',
-    from: 'claude-sonnet-4-5-20250929',
-    to: 'claude-sonnet-4-6',
-    color: 'bg-cyan-100 text-cyan-700 hover:bg-cyan-200 dark:bg-cyan-900/30 dark:text-cyan-400'
-  },
-  {
-    label: 'Sonnet3.5→4.6',
-    from: 'claude-3-5-sonnet-20241022',
-    to: 'claude-sonnet-4-6',
-    color: 'bg-teal-100 text-teal-700 hover:bg-teal-200 dark:bg-teal-900/30 dark:text-teal-400'
-  },
-  {
-    label: 'Opus4.5→4.6',
-    from: 'claude-opus-4-5-20251101',
-    to: 'claude-opus-4-6-thinking',
-    color:
-      'bg-violet-100 text-violet-700 hover:bg-violet-200 dark:bg-violet-900/30 dark:text-violet-400'
-  },
-  {
-    label: 'Opus->Sonnet',
-    from: 'claude-opus-4-5-20251101',
-    to: 'claude-sonnet-4-5-20250929',
-    color: 'bg-amber-100 text-amber-700 hover:bg-amber-200 dark:bg-amber-900/30 dark:text-amber-400'
-  },
-  {
-    label: 'Gemini 2.5 Image',
-    from: 'gemini-2.5-flash-image',
-    to: 'gemini-2.5-flash-image',
-    color: 'bg-sky-100 text-sky-700 hover:bg-sky-200 dark:bg-sky-900/30 dark:text-sky-400'
-  },
-  {
-    label: 'Gemini 3.1 Image',
-    from: 'gemini-3.1-flash-image',
-    to: 'gemini-3.1-flash-image',
-    color: 'bg-sky-100 text-sky-700 hover:bg-sky-200 dark:bg-sky-900/30 dark:text-sky-400'
-  },
-  {
-    label: 'G3 Image→3.1',
-    from: 'gemini-3-pro-image',
-    to: 'gemini-3.1-flash-image',
-    color: 'bg-sky-100 text-sky-700 hover:bg-sky-200 dark:bg-sky-900/30 dark:text-sky-400'
-  },
-  {
-    label: 'GPT-5.3 Codex',
-    from: 'gpt-5.3-codex',
-    to: 'gpt-5.3-codex',
-    color: 'bg-emerald-100 text-emerald-700 hover:bg-emerald-200 dark:bg-emerald-900/30 dark:text-emerald-400'
-  },
-  {
-    label: 'GPT-5.3 Spark',
-    from: 'gpt-5.3-codex-spark',
-    to: 'gpt-5.3-codex-spark',
-    color: 'bg-emerald-100 text-emerald-700 hover:bg-emerald-200 dark:bg-emerald-900/30 dark:text-emerald-400'
-  },
-  {
-    label: 'GPT-5.4',
-    from: 'gpt-5.4',
-    to: 'gpt-5.4',
-    color: 'bg-rose-100 text-rose-700 hover:bg-rose-200 dark:bg-rose-900/30 dark:text-rose-400'
-  },
-  {
-    label: '5.2→5.3',
-    from: 'gpt-5.2-codex',
-    to: 'gpt-5.3-codex',
-    color: 'bg-lime-100 text-lime-700 hover:bg-lime-200 dark:bg-lime-900/30 dark:text-lime-400'
-  },
-  {
-    label: 'GPT-5.2',
-    from: 'gpt-5.2-2025-12-11',
-    to: 'gpt-5.2-2025-12-11',
-    color: 'bg-green-100 text-green-700 hover:bg-green-200 dark:bg-green-900/30 dark:text-green-400'
-  },
-  {
-    label: 'GPT-5.2 Codex',
-    from: 'gpt-5.2-codex',
-    to: 'gpt-5.2-codex',
-    color: 'bg-blue-100 text-blue-700 hover:bg-blue-200 dark:bg-blue-900/30 dark:text-blue-400'
-  },
-  {
-    label: 'Max->Codex',
-    from: 'gpt-5.1-codex-max',
-    to: 'gpt-5.1-codex',
-    color: 'bg-pink-100 text-pink-700 hover:bg-pink-200 dark:bg-pink-900/30 dark:text-pink-400'
-  },
-  {
-    label: '3-Pro-Preview→3.1-Pro-High',
-    from: 'gemini-3-pro-preview',
-    to: 'gemini-3.1-pro-high',
-    color: 'bg-amber-100 text-amber-700 hover:bg-amber-200 dark:bg-amber-900/30 dark:text-amber-400'
-  },
-  {
-    label: '3-Pro-High→3.1-Pro-High',
-    from: 'gemini-3-pro-high',
-    to: 'gemini-3.1-pro-high',
-    color: 'bg-orange-100 text-orange-700 hover:bg-orange-200 dark:bg-orange-900/30 dark:text-orange-400'
-  },
-  {
-    label: '3-Pro-Low→3.1-Pro-Low',
-    from: 'gemini-3-pro-low',
-    to: 'gemini-3.1-pro-low',
-    color: 'bg-yellow-100 text-yellow-700 hover:bg-yellow-200 dark:bg-yellow-900/30 dark:text-yellow-400'
-  },
-  {
-    label: '3-Flash透传',
-    from: 'gemini-3-flash',
-    to: 'gemini-3-flash',
-    color: 'bg-lime-100 text-lime-700 hover:bg-lime-200 dark:bg-lime-900/30 dark:text-lime-400'
-  },
-  {
-    label: '2.5-Flash-Lite透传',
-    from: 'gemini-2.5-flash-lite',
-    to: 'gemini-2.5-flash-lite',
-    color: 'bg-green-100 text-green-700 hover:bg-green-200 dark:bg-green-900/30 dark:text-green-400'
-  }
-]
-
 // Common HTTP error codes
 const commonErrorCodes = [
   { value: 401, label: 'Unauthorized' },

frontend/src/components/account/ModelWhitelistSelector.vue

@@ -131,7 +131,8 @@ const { t } = useI18n()
 
 const props = defineProps<{
   modelValue: string[]
-  platform: string
+  platform?: string
+  platforms?: string[]
 }>()
 
 const emit = defineEmits<{
@@ -144,11 +145,36 @@ const showDropdown = ref(false)
 const searchQuery = ref('')
 const customModel = ref('')
 const isComposing = ref(false)
+const normalizedPlatforms = computed(() => {
+  const rawPlatforms =
+    props.platforms && props.platforms.length > 0
+      ? props.platforms
+      : props.platform
+        ? [props.platform]
+        : []
+
+  return Array.from(
+    new Set(
+      rawPlatforms
+        .map(platform => platform?.trim())
+        .filter((platform): platform is string => Boolean(platform))
+    )
+  )
+})
+
 const availableOptions = computed(() => {
-  if (props.platform === 'sora') {
-    return getModelsByPlatform('sora').map(m => ({ value: m, label: m }))
+  if (normalizedPlatforms.value.length === 0) {
+    return allModels
   }
-  return allModels
+
+  const allowedModels = new Set<string>()
+  for (const platform of normalizedPlatforms.value) {
+    for (const model of getModelsByPlatform(platform)) {
+      allowedModels.add(model)
+    }
+  }
+
+  return allModels.filter(model => allowedModels.has(model.value))
 })
 
 const filteredModels = computed(() => {
@@ -192,10 +218,13 @@ const handleEnter = () => {
 }
 
 const fillRelated = () => {
-  const models = getModelsByPlatform(props.platform)
   const newModels = [...props.modelValue]
-  for (const model of models) {
-    if (!newModels.includes(model)) newModels.push(model)
+  for (const platform of normalizedPlatforms.value) {
+    for (const model of getModelsByPlatform(platform)) {
+      if (!newModels.includes(model)) {
+        newModels.push(model)
+      }
+    }
   }
   emit('update:modelValue', newModels)
 }

frontend/src/components/account/UsageProgressBar.vue

@@ -1,5 +1,29 @@
 <template>
   <div>
+    <!-- Window stats row (above progress bar) -->
+    <div
+      v-if="windowStats"
+      class="mb-0.5 flex items-center"
+    >
+      <div class="flex items-center gap-1.5 text-[9px] text-gray-500 dark:text-gray-400">
+        <span class="rounded bg-gray-100 px-1.5 py-0.5 dark:bg-gray-800">
+          {{ formatRequests }} req
+        </span>
+        <span class="rounded bg-gray-100 px-1.5 py-0.5 dark:bg-gray-800">
+          {{ formatTokens }}
+        </span>
+        <span class="rounded bg-gray-100 px-1.5 py-0.5 dark:bg-gray-800">
+          A ${{ formatAccountCost }}
+        </span>
+        <span
+          v-if="windowStats?.user_cost != null"
+          class="rounded bg-gray-100 px-1.5 py-0.5 dark:bg-gray-800"
+        >
+          U ${{ formatUserCost }}
+        </span>
+      </div>
+    </div>
+
     <!-- Progress bar row -->
     <div class="flex items-center gap-1">
       <!-- Label badge (fixed width for alignment) -->
@@ -108,4 +132,32 @@ const formatResetTime = computed(() => {
   }
 })
 
+// Window stats formatters
+const formatRequests = computed(() => {
+  if (!props.windowStats) return ''
+  const r = props.windowStats.requests
+  if (r >= 1000000) return `${(r / 1000000).toFixed(1)}M`
+  if (r >= 1000) return `${(r / 1000).toFixed(1)}K`
+  return r.toString()
+})
+
+const formatTokens = computed(() => {
+  if (!props.windowStats) return ''
+  const t = props.windowStats.tokens
+  if (t >= 1000000000) return `${(t / 1000000000).toFixed(1)}B`
+  if (t >= 1000000) return `${(t / 1000000).toFixed(1)}M`
+  if (t >= 1000) return `${(t / 1000).toFixed(1)}K`
+  return t.toString()
+})
+
+const formatAccountCost = computed(() => {
+  if (!props.windowStats) return '0.00'
+  return props.windowStats.cost.toFixed(2)
+})
+
+const formatUserCost = computed(() => {
+  if (!props.windowStats || props.windowStats.user_cost == null) return '0.00'
+  return props.windowStats.user_cost.toFixed(2)
+})
+
 </script>

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

@@ -410,6 +410,18 @@
 
         <!-- Model Distribution -->
         <ModelDistributionChart :model-stats="stats.models" :loading="false" />
+
+        <EndpointDistributionChart
+          :endpoint-stats="stats.endpoints || []"
+          :loading="false"
+          :title="t('usage.inboundEndpoint')"
+        />
+
+        <EndpointDistributionChart
+          :endpoint-stats="stats.upstream_endpoints || []"
+          :loading="false"
+          :title="t('usage.upstreamEndpoint')"
+        />
       </template>
 
       <!-- No Data State -->
@@ -453,6 +465,7 @@ import { Line } from 'vue-chartjs'
 import BaseDialog from '@/components/common/BaseDialog.vue'
 import LoadingSpinner from '@/components/common/LoadingSpinner.vue'
 import ModelDistributionChart from '@/components/charts/ModelDistributionChart.vue'
+import EndpointDistributionChart from '@/components/charts/EndpointDistributionChart.vue'
 import Icon from '@/components/icons/Icon.vue'
 import { adminAPI } from '@/api/admin'
 import type { Account, AccountUsageStatsResponse } from '@/types'

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

@@ -24,7 +24,7 @@ const updateType = (value: string | number | boolean | null) => { emit('update:f
 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' }, { value: 'sora', label: 'Sora' }])
-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 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') }, { value: 'bedrock', label: 'AWS Bedrock' }])
 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') }, { value: 'temp_unschedulable', label: t('admin.accounts.status.tempUnschedulable') }])
 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

@@ -35,6 +35,19 @@
           </span>
         </template>
 
+        <template #cell-endpoint="{ row }">
+          <div class="max-w-[320px] space-y-1 text-xs">
+            <div class="break-all text-gray-700 dark:text-gray-300">
+              <span class="font-medium text-gray-500 dark:text-gray-400">{{ t('usage.inbound') }}:</span>
+              <span class="ml-1">{{ row.inbound_endpoint?.trim() || '-' }}</span>
+            </div>
+            <div class="break-all text-gray-700 dark:text-gray-300">
+              <span class="font-medium text-gray-500 dark:text-gray-400">{{ t('usage.upstream') }}:</span>
+              <span class="ml-1">{{ row.upstream_endpoint?.trim() || '-' }}</span>
+            </div>
+          </div>
+        </template>
+
         <template #cell-group="{ row }">
           <span v-if="row.group" class="inline-flex items-center rounded px-2 py-0.5 text-xs font-medium bg-indigo-100 text-indigo-800 dark:bg-indigo-900 dark:text-indigo-200">
             {{ row.group.name }}
@@ -328,6 +341,7 @@ const getRequestTypeBadgeClass = (row: AdminUsageLog): string => {
   if (requestType === 'sync') return 'bg-gray-100 text-gray-800 dark:bg-gray-700 dark:text-gray-200'
   return 'bg-amber-100 text-amber-800 dark:bg-amber-900 dark:text-amber-200'
 }
+
 const formatCacheTokens = (tokens: number): string => {
   if (tokens >= 1000000) return `${(tokens / 1000000).toFixed(1)}M`
   if (tokens >= 1000) return `${(tokens / 1000).toFixed(1)}K`

frontend/src/components/charts/EndpointDistributionChart.vue

@@ -0,0 +1,257 @@
+<template>
+  <div class="card p-4">
+    <div class="mb-4 flex items-start justify-between gap-3">
+      <h3 class="text-sm font-semibold text-gray-900 dark:text-white">
+        {{ title || t('usage.endpointDistribution') }}
+      </h3>
+      <div class="flex flex-col items-end gap-2">
+        <div
+          v-if="showSourceToggle"
+          class="inline-flex rounded-lg border border-gray-200 bg-gray-50 p-0.5 dark:border-gray-700 dark:bg-dark-800"
+        >
+          <button
+            type="button"
+            class="rounded-md px-2.5 py-1 text-xs font-medium transition-colors"
+            :class="source === 'inbound'
+              ? 'bg-white text-gray-900 shadow-sm dark:bg-dark-700 dark:text-white'
+              : 'text-gray-500 hover:text-gray-700 dark:text-gray-400 dark:hover:text-gray-200'"
+            @click="emit('update:source', 'inbound')"
+          >
+            {{ t('usage.inbound') }}
+          </button>
+          <button
+            type="button"
+            class="rounded-md px-2.5 py-1 text-xs font-medium transition-colors"
+            :class="source === 'upstream'
+              ? 'bg-white text-gray-900 shadow-sm dark:bg-dark-700 dark:text-white'
+              : 'text-gray-500 hover:text-gray-700 dark:text-gray-400 dark:hover:text-gray-200'"
+            @click="emit('update:source', 'upstream')"
+          >
+            {{ t('usage.upstream') }}
+          </button>
+          <button
+            type="button"
+            class="rounded-md px-2.5 py-1 text-xs font-medium transition-colors"
+            :class="source === 'path'
+              ? 'bg-white text-gray-900 shadow-sm dark:bg-dark-700 dark:text-white'
+              : 'text-gray-500 hover:text-gray-700 dark:text-gray-400 dark:hover:text-gray-200'"
+            @click="emit('update:source', 'path')"
+          >
+            {{ t('usage.path') }}
+          </button>
+        </div>
+
+        <div
+          v-if="showMetricToggle"
+          class="inline-flex rounded-lg border border-gray-200 bg-gray-50 p-0.5 dark:border-gray-700 dark:bg-dark-800"
+        >
+          <button
+            type="button"
+            class="rounded-md px-2.5 py-1 text-xs font-medium transition-colors"
+            :class="metric === 'tokens'
+              ? 'bg-white text-gray-900 shadow-sm dark:bg-dark-700 dark:text-white'
+              : 'text-gray-500 hover:text-gray-700 dark:text-gray-400 dark:hover:text-gray-200'"
+            @click="emit('update:metric', 'tokens')"
+          >
+            {{ t('admin.dashboard.metricTokens') }}
+          </button>
+          <button
+            type="button"
+            class="rounded-md px-2.5 py-1 text-xs font-medium transition-colors"
+            :class="metric === 'actual_cost'
+              ? 'bg-white text-gray-900 shadow-sm dark:bg-dark-700 dark:text-white'
+              : 'text-gray-500 hover:text-gray-700 dark:text-gray-400 dark:hover:text-gray-200'"
+            @click="emit('update:metric', 'actual_cost')"
+          >
+            {{ t('admin.dashboard.metricActualCost') }}
+          </button>
+        </div>
+      </div>
+    </div>
+    <div v-if="loading" class="flex h-48 items-center justify-center">
+      <LoadingSpinner />
+    </div>
+    <div v-else-if="displayEndpointStats.length > 0 && chartData" class="flex items-center gap-6">
+      <div class="h-48 w-48">
+        <Doughnut :data="chartData" :options="doughnutOptions" />
+      </div>
+      <div class="max-h-48 flex-1 overflow-y-auto">
+        <table class="w-full text-xs">
+          <thead>
+            <tr class="text-gray-500 dark:text-gray-400">
+              <th class="pb-2 text-left">{{ t('usage.endpoint') }}</th>
+              <th class="pb-2 text-right">{{ t('admin.dashboard.requests') }}</th>
+              <th class="pb-2 text-right">{{ t('admin.dashboard.tokens') }}</th>
+              <th class="pb-2 text-right">{{ t('admin.dashboard.actual') }}</th>
+              <th class="pb-2 text-right">{{ t('admin.dashboard.standard') }}</th>
+            </tr>
+          </thead>
+          <tbody>
+            <tr
+              v-for="item in displayEndpointStats"
+              :key="item.endpoint"
+              class="border-t border-gray-100 dark:border-gray-700"
+            >
+              <td class="max-w-[180px] truncate py-1.5 font-medium text-gray-900 dark:text-white" :title="item.endpoint">
+                {{ item.endpoint }}
+              </td>
+              <td class="py-1.5 text-right text-gray-600 dark:text-gray-400">
+                {{ formatNumber(item.requests) }}
+              </td>
+              <td class="py-1.5 text-right text-gray-600 dark:text-gray-400">
+                {{ formatTokens(item.total_tokens) }}
+              </td>
+              <td class="py-1.5 text-right text-green-600 dark:text-green-400">
+                ${{ formatCost(item.actual_cost) }}
+              </td>
+              <td class="py-1.5 text-right text-gray-400 dark:text-gray-500">
+                ${{ formatCost(item.cost) }}
+              </td>
+            </tr>
+          </tbody>
+        </table>
+      </div>
+    </div>
+    <div v-else class="flex h-48 items-center justify-center text-sm text-gray-500 dark:text-gray-400">
+      {{ t('admin.dashboard.noDataAvailable') }}
+    </div>
+  </div>
+</template>
+
+<script setup lang="ts">
+import { computed } from 'vue'
+import { useI18n } from 'vue-i18n'
+import { Chart as ChartJS, ArcElement, Tooltip, Legend } from 'chart.js'
+import { Doughnut } from 'vue-chartjs'
+import LoadingSpinner from '@/components/common/LoadingSpinner.vue'
+import type { EndpointStat } from '@/types'
+
+ChartJS.register(ArcElement, Tooltip, Legend)
+
+const { t } = useI18n()
+
+type DistributionMetric = 'tokens' | 'actual_cost'
+type EndpointSource = 'inbound' | 'upstream' | 'path'
+
+const props = withDefaults(
+  defineProps<{
+    endpointStats: EndpointStat[]
+    upstreamEndpointStats?: EndpointStat[]
+    endpointPathStats?: EndpointStat[]
+    loading?: boolean
+    title?: string
+    metric?: DistributionMetric
+    source?: EndpointSource
+    showMetricToggle?: boolean
+    showSourceToggle?: boolean
+  }>(),
+  {
+    upstreamEndpointStats: () => [],
+    endpointPathStats: () => [],
+    loading: false,
+    title: '',
+    metric: 'tokens',
+    source: 'inbound',
+    showMetricToggle: false,
+    showSourceToggle: false
+  }
+)
+
+const emit = defineEmits<{
+  'update:metric': [value: DistributionMetric]
+  'update:source': [value: EndpointSource]
+}>()
+
+const chartColors = [
+  '#3b82f6',
+  '#10b981',
+  '#f59e0b',
+  '#ef4444',
+  '#8b5cf6',
+  '#ec4899',
+  '#14b8a6',
+  '#f97316',
+  '#6366f1',
+  '#84cc16',
+  '#06b6d4',
+  '#a855f7'
+]
+
+const displayEndpointStats = computed(() => {
+  const sourceStats = props.source === 'upstream'
+    ? props.upstreamEndpointStats
+    : props.source === 'path'
+      ? props.endpointPathStats
+      : props.endpointStats
+  if (!sourceStats?.length) return []
+
+  const metricKey = props.metric === 'actual_cost' ? 'actual_cost' : 'total_tokens'
+  return [...sourceStats].sort((a, b) => b[metricKey] - a[metricKey])
+})
+
+const chartData = computed(() => {
+  if (!displayEndpointStats.value?.length) return null
+
+  return {
+    labels: displayEndpointStats.value.map((item) => item.endpoint),
+    datasets: [
+      {
+        data: displayEndpointStats.value.map((item) =>
+          props.metric === 'actual_cost' ? item.actual_cost : item.total_tokens
+        ),
+        backgroundColor: chartColors.slice(0, displayEndpointStats.value.length),
+        borderWidth: 0
+      }
+    ]
+  }
+})
+
+const doughnutOptions = computed(() => ({
+  responsive: true,
+  maintainAspectRatio: false,
+  plugins: {
+    legend: {
+      display: false
+    },
+    tooltip: {
+      callbacks: {
+        label: (context: any) => {
+          const value = context.raw as number
+          const total = context.dataset.data.reduce((a: number, b: number) => a + b, 0)
+          const percentage = total > 0 ? ((value / total) * 100).toFixed(1) : '0.0'
+          const formattedValue = props.metric === 'actual_cost'
+            ? `$${formatCost(value)}`
+            : formatTokens(value)
+          return `${context.label}: ${formattedValue} (${percentage}%)`
+        }
+      }
+    }
+  }
+}))
+
+const formatTokens = (value: number): string => {
+  if (value >= 1_000_000_000) {
+    return `${(value / 1_000_000_000).toFixed(2)}B`
+  } else if (value >= 1_000_000) {
+    return `${(value / 1_000_000).toFixed(2)}M`
+  } else if (value >= 1_000) {
+    return `${(value / 1_000).toFixed(2)}K`
+  }
+  return value.toLocaleString()
+}
+
+const formatNumber = (value: number): string => {
+  return value.toLocaleString()
+}
+
+const formatCost = (value: number): string => {
+  if (value >= 1000) {
+    return (value / 1000).toFixed(2) + 'K'
+  } else if (value >= 1) {
+    return value.toFixed(2)
+  } else if (value >= 0.01) {
+    return value.toFixed(3)
+  }
+  return value.toFixed(4)
+}
+</script>

frontend/src/i18n/locales/en.ts

@@ -718,6 +718,13 @@ export default {
     preparingExport: 'Preparing export...',
     model: 'Model',
     reasoningEffort: 'Reasoning Effort',
+    endpoint: 'Endpoint',
+    endpointDistribution: 'Endpoint Distribution',
+    inbound: 'Inbound',
+    upstream: 'Upstream',
+    path: 'Path',
+    inboundEndpoint: 'Inbound Endpoint',
+    upstreamEndpoint: 'Upstream Endpoint',
     type: 'Type',
     tokens: 'Tokens',
     cost: 'Cost',
@@ -3835,6 +3842,8 @@ export default {
         ignoreNoAvailableAccountsHint: 'When enabled, "No available accounts" errors will not be written to the error log (not recommended; usually a config issue).',
         ignoreInvalidApiKeyErrors: 'Ignore invalid API key errors',
         ignoreInvalidApiKeyErrorsHint: 'When enabled, invalid or missing API key errors (INVALID_API_KEY, API_KEY_REQUIRED) will not be written to the error log.',
+        ignoreInsufficientBalanceErrors: 'Ignore Insufficient Balance Errors',
+        ignoreInsufficientBalanceErrorsHint: 'When enabled, insufficient account balance errors will not be written to the error log.',
         autoRefresh: 'Auto Refresh',
         enableAutoRefresh: 'Enable auto refresh',
         enableAutoRefreshHint: 'Automatically refresh dashboard data at a fixed interval.',
@@ -3959,6 +3968,9 @@ export default {
         invitationCodeHint: 'When enabled, users must enter a valid invitation code to register',
         passwordReset: 'Password Reset',
         passwordResetHint: 'Allow users to reset their password via email',
+        frontendUrl: 'Frontend URL',
+        frontendUrlPlaceholder: 'https://example.com',
+        frontendUrlHint: 'Used to generate password reset links in emails. Example: https://example.com',
         totp: 'Two-Factor Authentication (2FA)',
         totpHint: 'Allow users to use authenticator apps like Google Authenticator',
         totpKeyNotConfigured:

frontend/src/i18n/locales/zh.ts

@@ -723,6 +723,13 @@ export default {
     preparingExport: '正在准备导出...',
     model: '模型',
     reasoningEffort: '推理强度',
+    endpoint: '端点',
+    endpointDistribution: '端点分布',
+    inbound: '入站',
+    upstream: '上游',
+    path: '路径',
+    inboundEndpoint: '入站端点',
+    upstreamEndpoint: '上游端点',
     type: '类型',
     tokens: 'Token',
     cost: '费用',
@@ -4009,6 +4016,8 @@ export default {
         ignoreNoAvailableAccountsHint: '启用后，"No available accounts" 错误将不会写入错误日志（不推荐，这通常是配置问题）。',
         ignoreInvalidApiKeyErrors: '忽略无效 API Key 错误',
         ignoreInvalidApiKeyErrorsHint: '启用后，无效或缺失 API Key 的错误（INVALID_API_KEY、API_KEY_REQUIRED）将不会写入错误日志。',
+        ignoreInsufficientBalanceErrors: '忽略余额不足错误',
+        ignoreInsufficientBalanceErrorsHint: '启用后，账号余额不足（Insufficient balance）的错误将不会写入错误日志。',
         autoRefresh: '自动刷新',
         enableAutoRefresh: '启用自动刷新',
         enableAutoRefreshHint: '自动刷新仪表板数据，启用后会定期拉取最新数据。',
@@ -4133,6 +4142,9 @@ export default {
         invitationCodeHint: '开启后，用户注册时需要填写有效的邀请码',
         passwordReset: '忘记密码',
         passwordResetHint: '允许用户通过邮箱重置密码',
+        frontendUrl: '前端地址',
+        frontendUrlPlaceholder: 'https://example.com',
+        frontendUrlHint: '用于生成邮件中的密码重置链接，例如 https://example.com',
         totp: '双因素认证 (2FA)',
         totpHint: '允许用户使用 Google Authenticator 等应用进行二次验证',
         totpKeyNotConfigured:

frontend/src/types/index.ts

@@ -962,6 +962,8 @@ export interface UsageLog {
   model: string
   service_tier?: string | null
   reasoning_effort?: string | null
+  inbound_endpoint?: string | null
+  upstream_endpoint?: string | null
 
   group_id: number | null
   subscription_id: number | null
@@ -1168,6 +1170,14 @@ export interface ModelStat {
   actual_cost: number // 实际扣除
 }
 
+export interface EndpointStat {
+  endpoint: string
+  requests: number
+  total_tokens: number
+  cost: number
+  actual_cost: number
+}
+
 export interface GroupStat {
   group_id: number
   group_name: string
@@ -1362,6 +1372,8 @@ export interface AccountUsageStatsResponse {
   history: AccountUsageHistory[]
   summary: AccountUsageSummary
   models: ModelStat[]
+  endpoints: EndpointStat[]
+  upstream_endpoints: EndpointStat[]
 }
 
 // ==================== User Attribute Types ====================

frontend/src/views/admin/GroupsView.vue

@@ -2368,6 +2368,23 @@ const closeCreateModal = () => {
   createModelRoutingRules.value = []
 }
 
+const normalizeOptionalLimit = (value: number | string | null | undefined): number | null => {
+  if (value === null || value === undefined) {
+    return null
+  }
+
+  if (typeof value === 'string') {
+    const trimmed = value.trim()
+    if (!trimmed) {
+      return null
+    }
+    const parsed = Number(trimmed)
+    return Number.isFinite(parsed) && parsed > 0 ? parsed : null
+  }
+
+  return Number.isFinite(value) && value > 0 ? value : null
+}
+
 const handleCreateGroup = async () => {
   if (!createForm.name.trim()) {
     appStore.showError(t('admin.groups.nameRequired'))
@@ -2379,9 +2396,17 @@ const handleCreateGroup = async () => {
     const { sora_storage_quota_gb: createQuotaGb, ...createRest } = createForm
     const requestData = {
       ...createRest,
+      daily_limit_usd: normalizeOptionalLimit(createForm.daily_limit_usd as number | string | null),
+      weekly_limit_usd: normalizeOptionalLimit(createForm.weekly_limit_usd as number | string | null),
+      monthly_limit_usd: normalizeOptionalLimit(createForm.monthly_limit_usd as number | string | null),
       sora_storage_quota_bytes: createQuotaGb ? Math.round(createQuotaGb * 1024 * 1024 * 1024) : 0,
       model_routing: convertRoutingRulesToApiFormat(createModelRoutingRules.value)
     }
+    // v-model.number 清空输入框时产生 ""，转为 null 让后端设为无限制
+    const emptyToNull = (v: any) => v === '' ? null : v
+    requestData.daily_limit_usd = emptyToNull(requestData.daily_limit_usd)
+    requestData.weekly_limit_usd = emptyToNull(requestData.weekly_limit_usd)
+    requestData.monthly_limit_usd = emptyToNull(requestData.monthly_limit_usd)
     await adminAPI.groups.create(requestData)
     appStore.showSuccess(t('admin.groups.groupCreated'))
     closeCreateModal()
@@ -2457,6 +2482,9 @@ const handleUpdateGroup = async () => {
     const { sora_storage_quota_gb: editQuotaGb, ...editRest } = editForm
     const payload = {
       ...editRest,
+      daily_limit_usd: normalizeOptionalLimit(editForm.daily_limit_usd as number | string | null),
+      weekly_limit_usd: normalizeOptionalLimit(editForm.weekly_limit_usd as number | string | null),
+      monthly_limit_usd: normalizeOptionalLimit(editForm.monthly_limit_usd as number | string | null),
       sora_storage_quota_bytes: editQuotaGb ? Math.round(editQuotaGb * 1024 * 1024 * 1024) : 0,
       fallback_group_id: editForm.fallback_group_id === null ? 0 : editForm.fallback_group_id,
       fallback_group_id_on_invalid_request:
@@ -2465,6 +2493,11 @@ const handleUpdateGroup = async () => {
           : editForm.fallback_group_id_on_invalid_request,
       model_routing: convertRoutingRulesToApiFormat(editModelRoutingRules.value)
     }
+    // v-model.number 清空输入框时产生 ""，转为 null 让后端设为无限制
+    const emptyToNull = (v: any) => v === '' ? null : v
+    payload.daily_limit_usd = emptyToNull(payload.daily_limit_usd)
+    payload.weekly_limit_usd = emptyToNull(payload.weekly_limit_usd)
+    payload.monthly_limit_usd = emptyToNull(payload.monthly_limit_usd)
     await adminAPI.groups.update(editingGroup.value.id, payload)
     appStore.showSuccess(t('admin.groups.groupUpdated'))
     closeEditModal()

frontend/src/views/admin/SettingsView.vue

@@ -653,6 +653,24 @@
               </div>
               <Toggle v-model="form.password_reset_enabled" />
             </div>
+            <!-- Frontend URL - Only show when password reset is enabled -->
+            <div
+              v-if="form.email_verify_enabled && form.password_reset_enabled"
+              class="border-t border-gray-100 pt-4 dark:border-dark-700"
+            >
+              <label class="mb-2 block text-sm font-medium text-gray-700 dark:text-gray-300">
+                {{ t('admin.settings.registration.frontendUrl') }}
+              </label>
+              <input
+                v-model="form.frontend_url"
+                type="url"
+                class="input"
+                :placeholder="t('admin.settings.registration.frontendUrlPlaceholder')"
+              />
+              <p class="mt-1.5 text-xs text-gray-500 dark:text-gray-400">
+                {{ t('admin.settings.registration.frontendUrlHint') }}
+              </p>
+            </div>
 
             <!-- TOTP 2FA -->
             <div
@@ -1586,6 +1604,7 @@
               </div>
               <Toggle v-model="form.smtp_use_tls" />
             </div>
+
           </div>
         </div>
 
@@ -1820,6 +1839,7 @@ const form = reactive<SettingsForm>({
   purchase_subscription_url: '',
   sora_client_enabled: false,
   custom_menu_items: [] as Array<{id: string; label: string; icon_svg: string; url: string; visibility: 'user' | 'admin'; sort_order: number}>,
+  frontend_url: '',
   smtp_host: '',
   smtp_port: 587,
   smtp_username: '',
@@ -2097,6 +2117,7 @@ async function saveSettings() {
       purchase_subscription_url: form.purchase_subscription_url,
       sora_client_enabled: form.sora_client_enabled,
       custom_menu_items: form.custom_menu_items,
+      frontend_url: form.frontend_url,
       smtp_host: form.smtp_host,
       smtp_port: form.smtp_port,
       smtp_username: form.smtp_username,

frontend/src/views/admin/UsageView.vue

@@ -26,7 +26,20 @@
             :show-metric-toggle="true"
           />
         </div>
-        <TokenUsageTrend :trend-data="trendData" :loading="chartsLoading" />
+        <div class="grid grid-cols-1 gap-6 lg:grid-cols-2">
+          <EndpointDistributionChart
+            v-model:source="endpointDistributionSource"
+            v-model:metric="endpointDistributionMetric"
+            :endpoint-stats="inboundEndpointStats"
+            :upstream-endpoint-stats="upstreamEndpointStats"
+            :endpoint-path-stats="endpointPathStats"
+            :loading="endpointStatsLoading"
+            :show-source-toggle="true"
+            :show-metric-toggle="true"
+            :title="t('usage.endpointDistribution')"
+          />
+          <TokenUsageTrend :trend-data="trendData" :loading="chartsLoading" />
+        </div>
       </div>
       <UsageFilters v-model="filters" v-model:startDate="startDate" v-model:endDate="endDate" :exporting="exporting" @change="applyFilters" @refresh="refreshData" @reset="resetFilters" @cleanup="openCleanupDialog" @export="exportToExcel">
         <template #after-reset>
@@ -99,19 +112,28 @@ import UsageTable from '@/components/admin/usage/UsageTable.vue'; import UsageEx
 import UsageCleanupDialog from '@/components/admin/usage/UsageCleanupDialog.vue'
 import UserBalanceHistoryModal from '@/components/admin/user/UserBalanceHistoryModal.vue'
 import ModelDistributionChart from '@/components/charts/ModelDistributionChart.vue'; import GroupDistributionChart from '@/components/charts/GroupDistributionChart.vue'; import TokenUsageTrend from '@/components/charts/TokenUsageTrend.vue'
+import EndpointDistributionChart from '@/components/charts/EndpointDistributionChart.vue'
 import Icon from '@/components/icons/Icon.vue'
-import type { AdminUsageLog, TrendDataPoint, ModelStat, GroupStat, AdminUser } from '@/types'; import type { AdminUsageStatsResponse, AdminUsageQueryParams } from '@/api/admin/usage'
+import type { AdminUsageLog, TrendDataPoint, ModelStat, GroupStat, EndpointStat, AdminUser } from '@/types'; import type { AdminUsageStatsResponse, AdminUsageQueryParams } from '@/api/admin/usage'
 
 const { t } = useI18n()
 const appStore = useAppStore()
 type DistributionMetric = 'tokens' | 'actual_cost'
+type EndpointSource = 'inbound' | 'upstream' | 'path'
 const route = useRoute()
 const usageStats = ref<AdminUsageStatsResponse | null>(null); const usageLogs = ref<AdminUsageLog[]>([]); const loading = ref(false); const exporting = ref(false)
 const trendData = ref<TrendDataPoint[]>([]); const modelStats = ref<ModelStat[]>([]); const groupStats = ref<GroupStat[]>([]); const chartsLoading = ref(false); const granularity = ref<'day' | 'hour'>('day')
 const modelDistributionMetric = ref<DistributionMetric>('tokens')
 const groupDistributionMetric = ref<DistributionMetric>('tokens')
+const endpointDistributionMetric = ref<DistributionMetric>('tokens')
+const endpointDistributionSource = ref<EndpointSource>('inbound')
+const inboundEndpointStats = ref<EndpointStat[]>([])
+const upstreamEndpointStats = ref<EndpointStat[]>([])
+const endpointPathStats = ref<EndpointStat[]>([])
+const endpointStatsLoading = ref(false)
 let abortController: AbortController | null = null; let exportAbortController: AbortController | null = null
 let chartReqSeq = 0
+let statsReqSeq = 0
 const exportProgress = reactive({ show: false, progress: 0, current: 0, total: 0, estimatedTime: '' })
 const cleanupDialogVisible = ref(false)
 // Balance history modal state
@@ -183,13 +205,25 @@ const loadLogs = async () => {
   } catch (error: any) { if(error?.name !== 'AbortError') console.error('Failed to load usage logs:', error) } finally { if(abortController === c) loading.value = false }
 }
 const loadStats = async () => {
+  const seq = ++statsReqSeq
+  endpointStatsLoading.value = true
   try {
     const requestType = filters.value.request_type
     const legacyStream = requestType ? requestTypeToLegacyStream(requestType) : filters.value.stream
     const s = await adminAPI.usage.getStats({ ...filters.value, stream: legacyStream === null ? undefined : legacyStream })
+    if (seq !== statsReqSeq) return
     usageStats.value = s
+    inboundEndpointStats.value = s.endpoints || []
+    upstreamEndpointStats.value = s.upstream_endpoints || []
+    endpointPathStats.value = s.endpoint_paths || []
   } catch (error) {
+    if (seq !== statsReqSeq) return
     console.error('Failed to load usage stats:', error)
+    inboundEndpointStats.value = []
+    upstreamEndpointStats.value = []
+    endpointPathStats.value = []
+  } finally {
+    if (seq === statsReqSeq) endpointStatsLoading.value = false
   }
 }
 const loadChartData = async () => {
@@ -246,6 +280,7 @@ const exportToExcel = async () => {
     const headers = [
       t('usage.time'), t('admin.usage.user'), t('usage.apiKeyFilter'),
       t('admin.usage.account'), t('usage.model'), t('usage.reasoningEffort'), t('admin.usage.group'),
+      t('usage.inboundEndpoint'), t('usage.upstreamEndpoint'),
       t('usage.type'),
       t('admin.usage.inputTokens'), t('admin.usage.outputTokens'),
       t('admin.usage.cacheReadTokens'), t('admin.usage.cacheCreationTokens'),
@@ -263,7 +298,8 @@ const exportToExcel = async () => {
       if (c.signal.aborted) break; if (p === 1) { total = res.total; exportProgress.total = total }
       const rows = (res.items || []).map((log: AdminUsageLog) => [
         log.created_at, log.user?.email || '', log.api_key?.name || '', log.account?.name || '', log.model,
-        formatReasoningEffort(log.reasoning_effort), log.group?.name || '', getRequestTypeLabel(log),
+        formatReasoningEffort(log.reasoning_effort), log.group?.name || '',
+        log.inbound_endpoint || '', log.upstream_endpoint || '', getRequestTypeLabel(log),
         log.input_tokens, log.output_tokens, log.cache_read_tokens, log.cache_creation_tokens,
         log.input_cost?.toFixed(6) || '0.000000', log.output_cost?.toFixed(6) || '0.000000',
         log.cache_read_cost?.toFixed(6) || '0.000000', log.cache_creation_cost?.toFixed(6) || '0.000000',
@@ -301,6 +337,7 @@ const allColumns = computed(() => [
   { key: 'account', label: t('admin.usage.account'), sortable: false },
   { key: 'model', label: t('usage.model'), sortable: true },
   { key: 'reasoning_effort', label: t('usage.reasoningEffort'), sortable: false },
+  { key: 'endpoint', label: t('usage.endpoint'), sortable: false },
   { key: 'group', label: t('admin.usage.group'), sortable: false },
   { key: 'stream', label: t('usage.type'), sortable: false },
   { key: 'tokens', label: t('usage.tokens'), sortable: false },
@@ -343,12 +380,18 @@ const loadSavedColumns = () => {
   try {
     const saved = localStorage.getItem(HIDDEN_COLUMNS_KEY)
     if (saved) {
-      (JSON.parse(saved) as string[]).forEach(key => hiddenColumns.add(key))
+      (JSON.parse(saved) as string[]).forEach((key) => {
+        hiddenColumns.add(key)
+      })
     } else {
-      DEFAULT_HIDDEN_COLUMNS.forEach(key => hiddenColumns.add(key))
+      DEFAULT_HIDDEN_COLUMNS.forEach((key) => {
+        hiddenColumns.add(key)
+      })
     }
   } catch {
-    DEFAULT_HIDDEN_COLUMNS.forEach(key => hiddenColumns.add(key))
+    DEFAULT_HIDDEN_COLUMNS.forEach((key) => {
+      hiddenColumns.add(key)
+    })
   }
 }
 

frontend/src/views/admin/ops/components/OpsSettingsDialog.vue

@@ -516,6 +516,16 @@ async function saveAllSettings() {
               </div>
               <Toggle v-model="advancedSettings.ignore_invalid_api_key_errors" />
             </div>
+
+            <div class="flex items-center justify-between">
+              <div>
+                <label class="text-sm font-medium text-gray-700 dark:text-gray-300">{{ t('admin.ops.settings.ignoreInsufficientBalanceErrors') }}</label>
+                <p class="mt-1 text-xs text-gray-500">
+                  {{ t('admin.ops.settings.ignoreInsufficientBalanceErrorsHint') }}
+                </p>
+              </div>
+              <Toggle v-model="advancedSettings.ignore_insufficient_balance_errors" />
+            </div>
           </div>
 
           <!-- Auto Refresh -->

frontend/src/views/user/UsageView.vue

@@ -166,6 +166,12 @@
             </span>
           </template>
 
+          <template #cell-endpoint="{ row }">
+            <span class="text-sm text-gray-600 dark:text-gray-300 block max-w-[320px] whitespace-normal break-all">
+              {{ formatUsageEndpoints(row) }}
+            </span>
+          </template>
+
           <template #cell-stream="{ row }">
             <span
               class="inline-flex items-center rounded px-2 py-0.5 text-xs font-medium"
@@ -516,6 +522,7 @@ const columns = computed<Column[]>(() => [
   { key: 'api_key', label: t('usage.apiKeyFilter'), sortable: false },
   { key: 'model', label: t('usage.model'), sortable: true },
   { key: 'reasoning_effort', label: t('usage.reasoningEffort'), sortable: false },
+  { key: 'endpoint', label: t('usage.endpoint'), sortable: false },
   { key: 'stream', label: t('usage.type'), sortable: false },
   { key: 'tokens', label: t('usage.tokens'), sortable: false },
   { key: 'cost', label: t('usage.cost'), sortable: false },
@@ -615,6 +622,11 @@ const getRequestTypeExportText = (log: UsageLog): string => {
   return 'Unknown'
 }
 
+const formatUsageEndpoints = (log: UsageLog): string => {
+  const inbound = log.inbound_endpoint?.trim()
+  return inbound || '-'
+}
+
 const formatTokens = (value: number): string => {
   if (value >= 1_000_000_000) {
     return `${(value / 1_000_000_000).toFixed(2)}B`
@@ -789,6 +801,7 @@ const exportToCSV = async () => {
       'API Key Name',
       'Model',
       'Reasoning Effort',
+      'Inbound Endpoint',
       'Type',
       'Input Tokens',
       'Output Tokens',
@@ -806,6 +819,7 @@ const exportToCSV = async () => {
         log.api_key?.name || '',
         log.model,
         formatReasoningEffort(log.reasoning_effort),
+        log.inbound_endpoint || '',
         getRequestTypeExportText(log),
         log.input_tokens,
         log.output_tokens,