chore: add more info for panic fix

百度千帆平台上的第三方大模型调用

README.md

@@ -67,6 +67,7 @@ _✨ 通过标准的 OpenAI API 格式访问所有的大模型，开箱即用
    + [x] [OpenAI ChatGPT 系列模型](https://platform.openai.com/docs/guides/gpt/chat-completions-api)（支持 [Azure OpenAI API](https://learn.microsoft.com/en-us/azure/ai-services/openai/reference)）
    + [x] [Anthropic Claude 系列模型](https://anthropic.com)
    + [x] [Google PaLM2/Gemini 系列模型](https://developers.generativeai.google)
+   + [x] [Mistral 系列模型](https://mistral.ai/)
    + [x] [百度文心一言系列模型](https://cloud.baidu.com/doc/WENXINWORKSHOP/index.html)
    + [x] [阿里通义千问系列模型](https://help.aliyun.com/document_detail/2400395.html)
    + [x] [讯飞星火认知大模型](https://www.xfyun.cn/doc/spark/Web.html)
@@ -77,6 +78,7 @@ _✨ 通过标准的 OpenAI API 格式访问所有的大模型，开箱即用
    + [x] [百川大模型](https://platform.baichuan-ai.com)
    + [ ] [字节云雀大模型](https://www.volcengine.com/product/ark) (WIP)
    + [x] [MINIMAX](https://api.minimax.chat/)
+   + [x] [Groq](https://wow.groq.com/)
 2. 支持配置镜像以及众多[第三方代理服务](https://iamazing.cn/page/openai-api-third-party-services)。
 3. 支持通过**负载均衡**的方式访问多个渠道。
 4. 支持 **stream 模式**，可以通过流式传输实现打字机效果。
@@ -104,6 +106,7 @@ _✨ 通过标准的 OpenAI API 格式访问所有的大模型，开箱即用
     + [GitHub 开放授权](https://github.com/settings/applications/new)。
     + 微信公众号授权（需要额外部署 [WeChat Server](https://github.com/songquanpeng/wechat-server)）。
 23. 支持主题切换，设置环境变量 `THEME` 即可，默认为 `default`，欢迎 PR 更多主题，具体参考[此处](./web/README.md)。
+24. 配合 [Message Pusher](https://github.com/songquanpeng/message-pusher) 可将报警信息推送到多种 App 上。
 
 ## 部署
 ### 基于 Docker 进行部署
@@ -373,6 +376,9 @@ graph LR
 16. `SQLITE_BUSY_TIMEOUT`：SQLite 锁等待超时设置，单位为毫秒，默认 `3000`。
 17. `GEMINI_SAFETY_SETTING`：Gemini 的安全设置，默认 `BLOCK_NONE`。
 18. `THEME`：系统的主题设置，默认为 `default`，具体可选值参考[此处](./web/README.md)。
+19. `ENABLE_METRIC`：是否根据请求成功率禁用渠道，默认不开启，可选值为 `true` 和 `false`。
+20. `METRIC_QUEUE_SIZE`：请求成功率统计队列大小，默认为 `10`。
+21. `METRIC_SUCCESS_RATE_THRESHOLD`：请求成功率阈值，默认为 `0.8`。
 
 ### 命令行参数
 1. `--port <port_number>`: 指定服务器监听的端口号，默认为 `3000`。

common/blacklist/main.go

@@ -0,0 +1,29 @@
+package blacklist
+
+import (
+	"fmt"
+	"sync"
+)
+
+var blackList sync.Map
+
+func init() {
+	blackList = sync.Map{}
+}
+
+func userId2Key(id int) string {
+	return fmt.Sprintf("userid_%d", id)
+}
+
+func BanUser(id int) {
+	blackList.Store(userId2Key(id), true)
+}
+
+func UnbanUser(id int) {
+	blackList.Delete(userId2Key(id))
+}
+
+func IsUserBanned(id int) bool {
+	_, ok := blackList.Load(userId2Key(id))
+	return ok
+}

common/config/config.go

@@ -52,6 +52,7 @@ var EmailDomainWhitelist = []string{
 }
 
 var DebugEnabled = os.Getenv("DEBUG") == "true"
+var DebugSQLEnabled = os.Getenv("DEBUG_SQL") == "true"
 var MemoryCacheEnabled = os.Getenv("MEMORY_CACHE_ENABLED") == "true"
 
 var LogConsumeEnabled = true
@@ -69,6 +70,9 @@ var WeChatServerAddress = ""
 var WeChatServerToken = ""
 var WeChatAccountQRCodeImageURL = ""
 
+var MessagePusherAddress = ""
+var MessagePusherToken = ""
+
 var TurnstileSiteKey = ""
 var TurnstileSecretKey = ""
 
@@ -125,3 +129,9 @@ var (
 )
 
 var RateLimitKeyExpirationDuration = 20 * time.Minute
+
+var EnableMetric = helper.GetOrDefaultEnvBool("ENABLE_METRIC", false)
+var MetricQueueSize = helper.GetOrDefaultEnvInt("METRIC_QUEUE_SIZE", 10)
+var MetricSuccessRateThreshold = helper.GetOrDefaultEnvFloat64("METRIC_SUCCESS_RATE_THRESHOLD", 0.8)
+var MetricSuccessChanSize = helper.GetOrDefaultEnvInt("METRIC_SUCCESS_CHAN_SIZE", 1024)
+var MetricFailChanSize = helper.GetOrDefaultEnvInt("METRIC_FAIL_CHAN_SIZE", 128)

common/constants.go

@@ -15,6 +15,7 @@ const (
 const (
 	UserStatusEnabled  = 1 // don't use 0, 0 is the default value!
 	UserStatusDisabled = 2 // also don't use 0
+	UserStatusDeleted  = 3
 )
 
 const (
@@ -38,34 +39,38 @@ const (
 )
 
 const (
-	ChannelTypeUnknown        = 0
+	ChannelTypeUnknown = iota
-	ChannelTypeOpenAI         = 1
+	ChannelTypeOpenAI
-	ChannelTypeAPI2D          = 2
+	ChannelTypeAPI2D
-	ChannelTypeAzure          = 3
+	ChannelTypeAzure
-	ChannelTypeCloseAI        = 4
+	ChannelTypeCloseAI
-	ChannelTypeOpenAISB       = 5
+	ChannelTypeOpenAISB
-	ChannelTypeOpenAIMax      = 6
+	ChannelTypeOpenAIMax
-	ChannelTypeOhMyGPT        = 7
+	ChannelTypeOhMyGPT
-	ChannelTypeCustom         = 8
+	ChannelTypeCustom
-	ChannelTypeAILS           = 9
+	ChannelTypeAILS
-	ChannelTypeAIProxy        = 10
+	ChannelTypeAIProxy
-	ChannelTypePaLM           = 11
+	ChannelTypePaLM
-	ChannelTypeAPI2GPT        = 12
+	ChannelTypeAPI2GPT
-	ChannelTypeAIGC2D         = 13
+	ChannelTypeAIGC2D
-	ChannelTypeAnthropic      = 14
+	ChannelTypeAnthropic
-	ChannelTypeBaidu          = 15
+	ChannelTypeBaidu
-	ChannelTypeZhipu          = 16
+	ChannelTypeZhipu
-	ChannelTypeAli            = 17
+	ChannelTypeAli
-	ChannelTypeXunfei         = 18
+	ChannelTypeXunfei
-	ChannelType360            = 19
+	ChannelType360
-	ChannelTypeOpenRouter     = 20
+	ChannelTypeOpenRouter
-	ChannelTypeAIProxyLibrary = 21
+	ChannelTypeAIProxyLibrary
-	ChannelTypeFastGPT        = 22
+	ChannelTypeFastGPT
-	ChannelTypeTencent        = 23
+	ChannelTypeTencent
-	ChannelTypeGemini         = 24
+	ChannelTypeGemini
-	ChannelTypeMoonshot       = 25
+	ChannelTypeMoonshot
-	ChannelTypeBaichuan       = 26
+	ChannelTypeBaichuan
-	ChannelTypeMinimax        = 27
+	ChannelTypeMinimax
+	ChannelTypeMistral
+	ChannelTypeGroq
+
+	ChannelTypeDummy
 )
 
 var ChannelBaseURLs = []string{
@@ -97,6 +102,8 @@ var ChannelBaseURLs = []string{
 	"https://api.moonshot.cn",                   // 25
 	"https://api.baichuan-ai.com",               // 26
 	"https://api.minimax.chat",                  // 27
+	"https://api.mistral.ai",                    // 28
+	"https://api.groq.com/openai",               // 29
 }
 
 const (

common/helper/helper.go

@@ -195,6 +195,13 @@ func Max(a int, b int) int {
 	}
 }
 
+func GetOrDefaultEnvBool(env string, defaultValue bool) bool {
+	if env == "" || os.Getenv(env) == "" {
+		return defaultValue
+	}
+	return os.Getenv(env) == "true"
+}
+
 func GetOrDefaultEnvInt(env string, defaultValue int) int {
 	if env == "" || os.Getenv(env) == "" {
 		return defaultValue
@@ -207,6 +214,18 @@ func GetOrDefaultEnvInt(env string, defaultValue int) int {
 	return num
 }
 
+func GetOrDefaultEnvFloat64(env string, defaultValue float64) float64 {
+	if env == "" || os.Getenv(env) == "" {
+		return defaultValue
+	}
+	num, err := strconv.ParseFloat(os.Getenv(env), 64)
+	if err != nil {
+		logger.SysError(fmt.Sprintf("failed to parse %s: %s, using default value: %f", env, err.Error(), defaultValue))
+		return defaultValue
+	}
+	return num
+}
+
 func GetOrDefaultEnvString(env string, defaultValue string) string {
 	if env == "" || os.Getenv(env) == "" {
 		return defaultValue

common/logger/logger.go

@@ -19,9 +19,6 @@ const (
 	loggerError = "ERR"
 )
 
-const maxLogCount = 1000000
-
-var logCount int
 var setupLogLock sync.Mutex
 var setupLogWorking bool
 
@@ -96,9 +93,7 @@ func logHelper(ctx context.Context, level string, msg string) {
 	id := ctx.Value(RequestIdKey)
 	now := time.Now()
 	_, _ = fmt.Fprintf(writer, "[%s] %v | %s | %s \n", level, now.Format("2006/01/02 - 15:04:05"), id, msg)
-	logCount++ // we don't need accurate count, so no lock here
+	if !setupLogWorking {
-	if logCount > maxLogCount && !setupLogWorking {
-		logCount = 0
 		setupLogWorking = true
 		go func() {
 			SetupLogger()

common/message/email.go

@@ -1,4 +1,4 @@
-package common
+package message
 
 import (
 	"crypto/rand"
@@ -12,6 +12,9 @@ import (
 )
 
 func SendEmail(subject string, receiver string, content string) error {
+	if receiver == "" {
+		return fmt.Errorf("receiver is empty")
+	}
 	if config.SMTPFrom == "" { // for compatibility
 		config.SMTPFrom = config.SMTPAccount
 	}

common/message/main.go

@@ -0,0 +1,22 @@
+package message
+
+import (
+	"fmt"
+	"github.com/songquanpeng/one-api/common/config"
+)
+
+const (
+	ByAll           = "all"
+	ByEmail         = "email"
+	ByMessagePusher = "message_pusher"
+)
+
+func Notify(by string, title string, description string, content string) error {
+	if by == ByEmail {
+		return SendEmail(title, config.RootUserEmail, content)
+	}
+	if by == ByMessagePusher {
+		return SendMessage(title, description, content)
+	}
+	return fmt.Errorf("unknown notify method: %s", by)
+}

common/message/message-pusher.go

@@ -0,0 +1,53 @@
+package message
+
+import (
+	"bytes"
+	"encoding/json"
+	"errors"
+	"github.com/songquanpeng/one-api/common/config"
+	"net/http"
+)
+
+type request struct {
+	Title       string `json:"title"`
+	Description string `json:"description"`
+	Content     string `json:"content"`
+	URL         string `json:"url"`
+	Channel     string `json:"channel"`
+	Token       string `json:"token"`
+}
+
+type response struct {
+	Success bool   `json:"success"`
+	Message string `json:"message"`
+}
+
+func SendMessage(title string, description string, content string) error {
+	if config.MessagePusherAddress == "" {
+		return errors.New("message pusher address is not set")
+	}
+	req := request{
+		Title:       title,
+		Description: description,
+		Content:     content,
+		Token:       config.MessagePusherToken,
+	}
+	data, err := json.Marshal(req)
+	if err != nil {
+		return err
+	}
+	resp, err := http.Post(config.MessagePusherAddress,
+		"application/json", bytes.NewBuffer(data))
+	if err != nil {
+		return err
+	}
+	var res response
+	err = json.NewDecoder(resp.Body).Decode(&res)
+	if err != nil {
+		return err
+	}
+	if !res.Success {
+		return errors.New(res.Message)
+	}
+	return nil
+}

common/model-ratio.go

@@ -7,29 +7,6 @@ import (
 	"time"
 )
 
-var DalleSizeRatios = map[string]map[string]float64{
-	"dall-e-2": {
-		"256x256":   1,
-		"512x512":   1.125,
-		"1024x1024": 1.25,
-	},
-	"dall-e-3": {
-		"1024x1024": 1,
-		"1024x1792": 2,
-		"1792x1024": 2,
-	},
-}
-
-var DalleGenerationImageAmounts = map[string][2]int{
-	"dall-e-2": {1, 10},
-	"dall-e-3": {1, 1}, // OpenAI allows n=1 currently.
-}
-
-var DalleImagePromptLengthLimitations = map[string]int{
-	"dall-e-2": 1000,
-	"dall-e-3": 4000,
-}
-
 const (
 	USD2RMB = 7
 	USD     = 500 // $0.002 = 1 -> $1 = 500
@@ -40,7 +17,6 @@ const (
 // https://platform.openai.com/docs/models/model-endpoint-compatibility
 // https://cloud.baidu.com/doc/WENXINWORKSHOP/s/Blfmc9dlf
 // https://openai.com/pricing
-// TODO: when a new api is enabled, check the pricing here
 // 1 === $0.002 / 1K tokens
 // 1 === ￥0.014 / 1k tokens
 var ModelRatio = map[string]float64{
@@ -87,12 +63,15 @@ var ModelRatio = map[string]float64{
 	"text-search-ada-doc-001": 10,
 	"text-moderation-stable":  0.1,
 	"text-moderation-latest":  0.1,
-	"dall-e-2":                8,     // $0.016 - $0.020 / image
+	"dall-e-2":                8,  // $0.016 - $0.020 / image
-	"dall-e-3":                20,    // $0.040 - $0.120 / image
+	"dall-e-3":                20, // $0.040 - $0.120 / image
-	"claude-instant-1":        0.815, // $1.63 / 1M tokens
+	// https://www.anthropic.com/api#pricing
-	"claude-2":                5.51,  // $11.02 / 1M tokens
+	"claude-instant-1.2":       0.8 / 1000 * USD,
-	"claude-2.0":              5.51,  // $11.02 / 1M tokens
+	"claude-2.0":               8.0 / 1000 * USD,
-	"claude-2.1":              5.51,  // $11.02 / 1M tokens
+	"claude-2.1":               8.0 / 1000 * USD,
+	"claude-3-haiku-20240229":  0.25 / 1000 * USD,
+	"claude-3-sonnet-20240229": 3.0 / 1000 * USD,
+	"claude-3-opus-20240229":   15.0 / 1000 * USD,
 	// https://cloud.baidu.com/doc/WENXINWORKSHOP/s/hlrk4akp7
 	"ERNIE-Bot":         0.8572,     // ￥0.012 / 1k tokens
 	"ERNIE-Bot-turbo":   0.5715,     // ￥0.008 / 1k tokens
@@ -139,15 +118,54 @@ var ModelRatio = map[string]float64{
 	"abab6-chat":    0.1 * RMB,
 	"abab5.5-chat":  0.015 * RMB,
 	"abab5.5s-chat": 0.005 * RMB,
+	// https://docs.mistral.ai/platform/pricing/
+	"open-mistral-7b":       0.25 / 1000 * USD,
+	"open-mixtral-8x7b":     0.7 / 1000 * USD,
+	"mistral-small-latest":  2.0 / 1000 * USD,
+	"mistral-medium-latest": 2.7 / 1000 * USD,
+	"mistral-large-latest":  8.0 / 1000 * USD,
+	"mistral-embed":         0.1 / 1000 * USD,
+	// https://wow.groq.com/
+	"llama2-70b-4096":    0.7 / 1000 * USD,
+	"llama2-7b-2048":     0.1 / 1000 * USD,
+	"mixtral-8x7b-32768": 0.27 / 1000 * USD,
+	"gemma-7b-it":        0.1 / 1000 * USD,
 }
 
+var CompletionRatio = map[string]float64{}
+
 var DefaultModelRatio map[string]float64
+var DefaultCompletionRatio map[string]float64
 
 func init() {
 	DefaultModelRatio = make(map[string]float64)
 	for k, v := range ModelRatio {
 		DefaultModelRatio[k] = v
 	}
+	DefaultCompletionRatio = make(map[string]float64)
+	for k, v := range CompletionRatio {
+		DefaultCompletionRatio[k] = v
+	}
+}
+
+func AddNewMissingRatio(oldRatio string) string {
+	newRatio := make(map[string]float64)
+	err := json.Unmarshal([]byte(oldRatio), &newRatio)
+	if err != nil {
+		logger.SysError("error unmarshalling old ratio: " + err.Error())
+		return oldRatio
+	}
+	for k, v := range DefaultModelRatio {
+		if _, ok := newRatio[k]; !ok {
+			newRatio[k] = v
+		}
+	}
+	jsonBytes, err := json.Marshal(newRatio)
+	if err != nil {
+		logger.SysError("error marshalling new ratio: " + err.Error())
+		return oldRatio
+	}
+	return string(jsonBytes)
 }
 
 func ModelRatio2JSONString() string {
@@ -178,8 +196,6 @@ func GetModelRatio(name string) float64 {
 	return ratio
 }
 
-var CompletionRatio = map[string]float64{}
-
 func CompletionRatio2JSONString() string {
 	jsonBytes, err := json.Marshal(CompletionRatio)
 	if err != nil {
@@ -197,6 +213,9 @@ func GetCompletionRatio(name string) float64 {
 	if ratio, ok := CompletionRatio[name]; ok {
 		return ratio
 	}
+	if ratio, ok := DefaultCompletionRatio[name]; ok {
+		return ratio
+	}
 	if strings.HasPrefix(name, "gpt-3.5") {
 		if strings.HasSuffix(name, "0125") {
 			// https://openai.com/blog/new-embedding-models-and-api-updates
@@ -215,7 +234,7 @@ func GetCompletionRatio(name string) float64 {
 				return 2
 			}
 		}
-		return 1.333333
+		return 4.0 / 3.0
 	}
 	if strings.HasPrefix(name, "gpt-4") {
 		if strings.HasSuffix(name, "preview") {
@@ -223,11 +242,18 @@ func GetCompletionRatio(name string) float64 {
 		}
 		return 2
 	}
-	if strings.HasPrefix(name, "claude-instant-1") {
+	if strings.HasPrefix(name, "claude-3") {
-		return 3.38
+		return 5
 	}
-	if strings.HasPrefix(name, "claude-2") {
+	if strings.HasPrefix(name, "claude-") {
-		return 2.965517
+		return 3
+	}
+	if strings.HasPrefix(name, "mistral-") {
+		return 3
+	}
+	switch name {
+	case "llama2-70b-4096":
+		return 0.8 / 0.7
 	}
 	return 1
 }

common/random.go

@@ -0,0 +1,8 @@
+package common
+
+import "math/rand"
+
+// RandRange returns a random number between min and max (max is not included)
+func RandRange(min, max int) int {
+	return min + rand.Intn(max-min)
+}

controller/channel-billing.go

@@ -8,6 +8,7 @@ import (
 	"github.com/songquanpeng/one-api/common/config"
 	"github.com/songquanpeng/one-api/common/logger"
 	"github.com/songquanpeng/one-api/model"
+	"github.com/songquanpeng/one-api/monitor"
 	"github.com/songquanpeng/one-api/relay/util"
 	"io"
 	"net/http"
@@ -295,7 +296,7 @@ func UpdateChannelBalance(c *gin.Context) {
 }
 
 func updateAllChannelsBalance() error {
-	channels, err := model.GetAllChannels(0, 0, true)
+	channels, err := model.GetAllChannels(0, 0, "all")
 	if err != nil {
 		return err
 	}
@@ -313,7 +314,7 @@ func updateAllChannelsBalance() error {
 		} else {
 			// err is nil & balance <= 0 means quota is used up
 			if balance <= 0 {
-				disableChannel(channel.Id, channel.Name, "余额不足")
+				monitor.DisableChannel(channel.Id, channel.Name, "余额不足")
 			}
 		}
 		time.Sleep(config.RequestInterval)
@@ -322,15 +323,14 @@ func updateAllChannelsBalance() error {
 }
 
 func UpdateAllChannelsBalance(c *gin.Context) {
-	// TODO: make it async
+	//err := updateAllChannelsBalance()
-	err := updateAllChannelsBalance()
+	//if err != nil {
-	if err != nil {
+	//	c.JSON(http.StatusOK, gin.H{
-		c.JSON(http.StatusOK, gin.H{
+	//		"success": false,
-			"success": false,
+	//		"message": err.Error(),
-			"message": err.Error(),
+	//	})
-		})
+	//	return
-		return
+	//}
-	}
 	c.JSON(http.StatusOK, gin.H{
 		"success": true,
 		"message": "",

controller/channel-test.go

@@ -8,7 +8,10 @@ import (
 	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/common/config"
 	"github.com/songquanpeng/one-api/common/logger"
+	"github.com/songquanpeng/one-api/common/message"
+	"github.com/songquanpeng/one-api/middleware"
 	"github.com/songquanpeng/one-api/model"
+	"github.com/songquanpeng/one-api/monitor"
 	"github.com/songquanpeng/one-api/relay/constant"
 	"github.com/songquanpeng/one-api/relay/helper"
 	relaymodel "github.com/songquanpeng/one-api/relay/model"
@@ -18,6 +21,7 @@ import (
 	"net/http/httptest"
 	"net/url"
 	"strconv"
+	"strings"
 	"sync"
 	"time"
 
@@ -51,6 +55,7 @@ func testChannel(channel *model.Channel) (err error, openaiErr *relaymodel.Error
 	c.Request.Header.Set("Content-Type", "application/json")
 	c.Set("channel", channel.Type)
 	c.Set("base_url", channel.GetBaseURL())
+	middleware.SetupContextForSelectedChannel(c, channel, "")
 	meta := util.GetRelayMeta(c)
 	apiType := constant.ChannelType2APIType(channel.Type)
 	adaptor := helper.GetAdaptor(apiType)
@@ -59,6 +64,12 @@ func testChannel(channel *model.Channel) (err error, openaiErr *relaymodel.Error
 	}
 	adaptor.Init(meta)
 	modelName := adaptor.GetModelList()[0]
+	if !strings.Contains(channel.Models, modelName) {
+		modelNames := strings.Split(channel.Models, ",")
+		if len(modelNames) > 0 {
+			modelName = modelNames[0]
+		}
+	}
 	request := buildTestRequest()
 	request.Model = modelName
 	meta.OriginModelName, meta.ActualModelName = modelName, modelName
@@ -139,33 +150,7 @@ func TestChannel(c *gin.Context) {
 var testAllChannelsLock sync.Mutex
 var testAllChannelsRunning bool = false
 
-func notifyRootUser(subject string, content string) {
+func testChannels(notify bool, scope string) error {
-	if config.RootUserEmail == "" {
-		config.RootUserEmail = model.GetRootUserEmail()
-	}
-	err := common.SendEmail(subject, config.RootUserEmail, content)
-	if err != nil {
-		logger.SysError(fmt.Sprintf("failed to send email: %s", err.Error()))
-	}
-}
-
-// disable & notify
-func disableChannel(channelId int, channelName string, reason string) {
-	model.UpdateChannelStatusById(channelId, common.ChannelStatusAutoDisabled)
-	subject := fmt.Sprintf("通道「%s」（#%d）已被禁用", channelName, channelId)
-	content := fmt.Sprintf("通道「%s」（#%d）已被禁用，原因：%s", channelName, channelId, reason)
-	notifyRootUser(subject, content)
-}
-
-// enable & notify
-func enableChannel(channelId int, channelName string) {
-	model.UpdateChannelStatusById(channelId, common.ChannelStatusEnabled)
-	subject := fmt.Sprintf("通道「%s」（#%d）已被启用", channelName, channelId)
-	content := fmt.Sprintf("通道「%s」（#%d）已被启用", channelName, channelId)
-	notifyRootUser(subject, content)
-}
-
-func testAllChannels(notify bool) error {
 	if config.RootUserEmail == "" {
 		config.RootUserEmail = model.GetRootUserEmail()
 	}
@@ -176,7 +161,7 @@ func testAllChannels(notify bool) error {
 	}
 	testAllChannelsRunning = true
 	testAllChannelsLock.Unlock()
-	channels, err := model.GetAllChannels(0, 0, true)
+	channels, err := model.GetAllChannels(0, 0, scope)
 	if err != nil {
 		return err
 	}
@@ -193,13 +178,17 @@ func testAllChannels(notify bool) error {
 			milliseconds := tok.Sub(tik).Milliseconds()
 			if isChannelEnabled && milliseconds > disableThreshold {
 				err = errors.New(fmt.Sprintf("响应时间 %.2fs 超过阈值 %.2fs", float64(milliseconds)/1000.0, float64(disableThreshold)/1000.0))
-				disableChannel(channel.Id, channel.Name, err.Error())
+				if config.AutomaticDisableChannelEnabled {
+					monitor.DisableChannel(channel.Id, channel.Name, err.Error())
+				} else {
+					_ = message.Notify(message.ByAll, fmt.Sprintf("渠道 %s （%d）测试超时", channel.Name, channel.Id), "", err.Error())
+				}
 			}
 			if isChannelEnabled && util.ShouldDisableChannel(openaiErr, -1) {
-				disableChannel(channel.Id, channel.Name, err.Error())
+				monitor.DisableChannel(channel.Id, channel.Name, err.Error())
 			}
 			if !isChannelEnabled && util.ShouldEnableChannel(err, openaiErr) {
-				enableChannel(channel.Id, channel.Name)
+				monitor.EnableChannel(channel.Id, channel.Name)
 			}
 			channel.UpdateResponseTime(milliseconds)
 			time.Sleep(config.RequestInterval)
@@ -208,7 +197,7 @@ func testAllChannels(notify bool) error {
 		testAllChannelsRunning = false
 		testAllChannelsLock.Unlock()
 		if notify {
-			err := common.SendEmail("通道测试完成", config.RootUserEmail, "通道测试完成，如果没有收到禁用通知，说明所有通道都正常")
+			err := message.Notify(message.ByAll, "通道测试完成", "", "通道测试完成，如果没有收到禁用通知，说明所有通道都正常")
 			if err != nil {
 				logger.SysError(fmt.Sprintf("failed to send email: %s", err.Error()))
 			}
@@ -217,8 +206,12 @@ func testAllChannels(notify bool) error {
 	return nil
 }
 
-func TestAllChannels(c *gin.Context) {
+func TestChannels(c *gin.Context) {
-	err := testAllChannels(true)
+	scope := c.Query("scope")
+	if scope == "" {
+		scope = "all"
+	}
+	err := testChannels(true, scope)
 	if err != nil {
 		c.JSON(http.StatusOK, gin.H{
 			"success": false,
@@ -237,7 +230,7 @@ func AutomaticallyTestChannels(frequency int) {
 	for {
 		time.Sleep(time.Duration(frequency) * time.Minute)
 		logger.SysLog("testing all channels")
-		_ = testAllChannels(false)
+		_ = testChannels(false, "all")
 		logger.SysLog("channel test finished")
 	}
 }

controller/channel.go

@@ -15,7 +15,7 @@ func GetAllChannels(c *gin.Context) {
 	if p < 0 {
 		p = 0
 	}
-	channels, err := model.GetAllChannels(p*config.ItemsPerPage, config.ItemsPerPage, false)
+	channels, err := model.GetAllChannels(p*config.ItemsPerPage, config.ItemsPerPage, "limited")
 	if err != nil {
 		c.JSON(http.StatusOK, gin.H{
 			"success": false,

controller/misc.go

@@ -5,6 +5,7 @@ import (
 	"fmt"
 	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/common/config"
+	"github.com/songquanpeng/one-api/common/message"
 	"github.com/songquanpeng/one-api/model"
 	"net/http"
 	"strings"
@@ -110,7 +111,7 @@ func SendEmailVerification(c *gin.Context) {
 	content := fmt.Sprintf("<p>您好，你正在进行%s邮箱验证。</p>"+
 		"<p>您的验证码为: <strong>%s</strong></p>"+
 		"<p>验证码 %d 分钟内有效，如果不是本人操作，请忽略。</p>", config.SystemName, code, common.VerificationValidMinutes)
-	err := common.SendEmail(subject, email, content)
+	err := message.SendEmail(subject, email, content)
 	if err != nil {
 		c.JSON(http.StatusOK, gin.H{
 			"success": false,
@@ -149,7 +150,7 @@ func SendPasswordResetEmail(c *gin.Context) {
 		"<p>点击 <a href='%s'>此处</a> 进行密码重置。</p>"+
 		"<p>如果链接无法点击，请尝试点击下面的链接或将其复制到浏览器中打开：<br> %s </p>"+
 		"<p>重置链接 %d 分钟内有效，如果不是本人操作，请忽略。</p>", config.SystemName, link, link, common.VerificationValidMinutes)
-	err := common.SendEmail(subject, email, content)
+	err := message.SendEmail(subject, email, content)
 	if err != nil {
 		c.JSON(http.StatusOK, gin.H{
 			"success": false,

controller/model.go

@@ -3,13 +3,13 @@ package controller
 import (
 	"fmt"
 	"github.com/gin-gonic/gin"
-	"github.com/songquanpeng/one-api/relay/channel/ai360"
+	"github.com/songquanpeng/one-api/common"
-	"github.com/songquanpeng/one-api/relay/channel/baichuan"
+	"github.com/songquanpeng/one-api/relay/channel/openai"
-	"github.com/songquanpeng/one-api/relay/channel/minimax"
-	"github.com/songquanpeng/one-api/relay/channel/moonshot"
 	"github.com/songquanpeng/one-api/relay/constant"
 	"github.com/songquanpeng/one-api/relay/helper"
 	relaymodel "github.com/songquanpeng/one-api/relay/model"
+	"github.com/songquanpeng/one-api/relay/util"
+	"net/http"
 )
 
 // https://platform.openai.com/docs/api-reference/models/list
@@ -41,6 +41,7 @@ type OpenAIModels struct {
 
 var openAIModels []OpenAIModels
 var openAIModelsMap map[string]OpenAIModels
+var channelId2Models map[int][]string
 
 func init() {
 	var permission []OpenAIModelPermission
@@ -78,54 +79,44 @@ func init() {
 			})
 		}
 	}
-	for _, modelName := range ai360.ModelList {
+	for _, channelType := range openai.CompatibleChannels {
-		openAIModels = append(openAIModels, OpenAIModels{
+		if channelType == common.ChannelTypeAzure {
-			Id:         modelName,
+			continue
-			Object:     "model",
+		}
-			Created:    1626777600,
+		channelName, channelModelList := openai.GetCompatibleChannelMeta(channelType)
-			OwnedBy:    "360",
+		for _, modelName := range channelModelList {
-			Permission: permission,
+			openAIModels = append(openAIModels, OpenAIModels{
-			Root:       modelName,
+				Id:         modelName,
-			Parent:     nil,
+				Object:     "model",
-		})
+				Created:    1626777600,
-	}
+				OwnedBy:    channelName,
-	for _, modelName := range moonshot.ModelList {
+				Permission: permission,
-		openAIModels = append(openAIModels, OpenAIModels{
+				Root:       modelName,
-			Id:         modelName,
+				Parent:     nil,
-			Object:     "model",
+			})
-			Created:    1626777600,
+		}
-			OwnedBy:    "moonshot",
-			Permission: permission,
-			Root:       modelName,
-			Parent:     nil,
-		})
-	}
-	for _, modelName := range baichuan.ModelList {
-		openAIModels = append(openAIModels, OpenAIModels{
-			Id:         modelName,
-			Object:     "model",
-			Created:    1626777600,
-			OwnedBy:    "baichuan",
-			Permission: permission,
-			Root:       modelName,
-			Parent:     nil,
-		})
-	}
-	for _, modelName := range minimax.ModelList {
-		openAIModels = append(openAIModels, OpenAIModels{
-			Id:         modelName,
-			Object:     "model",
-			Created:    1626777600,
-			OwnedBy:    "minimax",
-			Permission: permission,
-			Root:       modelName,
-			Parent:     nil,
-		})
 	}
 	openAIModelsMap = make(map[string]OpenAIModels)
 	for _, model := range openAIModels {
 		openAIModelsMap[model.Id] = model
 	}
+	channelId2Models = make(map[int][]string)
+	for i := 1; i < common.ChannelTypeDummy; i++ {
+		adaptor := helper.GetAdaptor(constant.ChannelType2APIType(i))
+		meta := &util.RelayMeta{
+			ChannelType: i,
+		}
+		adaptor.Init(meta)
+		channelId2Models[i] = adaptor.GetModelList()
+	}
+}
+
+func DashboardListModels(c *gin.Context) {
+	c.JSON(http.StatusOK, gin.H{
+		"success": true,
+		"message": "",
+		"data":    channelId2Models,
+	})
 }
 
 func ListModels(c *gin.Context) {

controller/relay.go

@@ -11,6 +11,7 @@ import (
 	"github.com/songquanpeng/one-api/common/logger"
 	"github.com/songquanpeng/one-api/middleware"
 	dbmodel "github.com/songquanpeng/one-api/model"
+	"github.com/songquanpeng/one-api/monitor"
 	"github.com/songquanpeng/one-api/relay/constant"
 	"github.com/songquanpeng/one-api/relay/controller"
 	"github.com/songquanpeng/one-api/relay/model"
@@ -45,11 +46,12 @@ func Relay(c *gin.Context) {
 		requestBody, _ := common.GetRequestBody(c)
 		logger.Debugf(ctx, "request body: %s", string(requestBody))
 	}
+	channelId := c.GetInt("channel_id")
 	bizErr := relay(c, relayMode)
 	if bizErr == nil {
+		monitor.Emit(channelId, true)
 		return
 	}
-	channelId := c.GetInt("channel_id")
 	lastFailedChannelId := channelId
 	channelName := c.GetString("channel_name")
 	group := c.GetString("group")
@@ -62,7 +64,7 @@ func Relay(c *gin.Context) {
 		retryTimes = 0
 	}
 	for i := retryTimes; i > 0; i-- {
-		channel, err := dbmodel.CacheGetRandomSatisfiedChannel(group, originalModel)
+		channel, err := dbmodel.CacheGetRandomSatisfiedChannel(group, originalModel, i != retryTimes)
 		if err != nil {
 			logger.Errorf(ctx, "CacheGetRandomSatisfiedChannel failed: %w", err)
 			break
@@ -117,7 +119,9 @@ func processChannelRelayError(ctx context.Context, channelId int, channelName st
 	logger.Errorf(ctx, "relay error (channel #%d): %s", channelId, err.Message)
 	// https://platform.openai.com/docs/guides/error-codes/api-errors
 	if util.ShouldDisableChannel(&err.Error, err.StatusCode) {
-		disableChannel(channelId, channelName, err.Message)
+		monitor.DisableChannel(channelId, channelName, err.Message)
+	} else {
+		monitor.Emit(channelId, false)
 	}
 }
 

main.go

@@ -64,13 +64,6 @@ func main() {
 		go model.SyncOptions(config.SyncFrequency)
 		go model.SyncChannelCache(config.SyncFrequency)
 	}
-	if os.Getenv("CHANNEL_UPDATE_FREQUENCY") != "" {
-		frequency, err := strconv.Atoi(os.Getenv("CHANNEL_UPDATE_FREQUENCY"))
-		if err != nil {
-			logger.FatalLog("failed to parse CHANNEL_UPDATE_FREQUENCY: " + err.Error())
-		}
-		go controller.AutomaticallyUpdateChannels(frequency)
-	}
 	if os.Getenv("CHANNEL_TEST_FREQUENCY") != "" {
 		frequency, err := strconv.Atoi(os.Getenv("CHANNEL_TEST_FREQUENCY"))
 		if err != nil {
@@ -83,6 +76,9 @@ func main() {
 		logger.SysLog("batch update enabled with interval " + strconv.Itoa(config.BatchUpdateInterval) + "s")
 		model.InitBatchUpdater()
 	}
+	if config.EnableMetric {
+		logger.SysLog("metric enabled, will disable channel if too much request failed")
+	}
 	openai.InitTokenEncoders()
 
 	// Initialize HTTP server

middleware/auth.go

@@ -4,6 +4,7 @@ import (
 	"github.com/gin-contrib/sessions"
 	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/common"
+	"github.com/songquanpeng/one-api/common/blacklist"
 	"github.com/songquanpeng/one-api/model"
 	"net/http"
 	"strings"
@@ -42,11 +43,14 @@ func authHelper(c *gin.Context, minRole int) {
 			return
 		}
 	}
-	if status.(int) == common.UserStatusDisabled {
+	if status.(int) == common.UserStatusDisabled || blacklist.IsUserBanned(id.(int)) {
 		c.JSON(http.StatusOK, gin.H{
 			"success": false,
 			"message": "用户已被封禁",
 		})
+		session := sessions.Default(c)
+		session.Clear()
+		_ = session.Save()
 		c.Abort()
 		return
 	}
@@ -99,7 +103,7 @@ func TokenAuth() func(c *gin.Context) {
 			abortWithMessage(c, http.StatusInternalServerError, err.Error())
 			return
 		}
-		if !userEnabled {
+		if !userEnabled || blacklist.IsUserBanned(token.UserId) {
 			abortWithMessage(c, http.StatusForbidden, "用户已被封禁")
 			return
 		}

middleware/distributor.go

@@ -68,7 +68,7 @@ func Distribute() func(c *gin.Context) {
 				}
 			}
 			requestModel = modelRequest.Model
-			channel, err = model.CacheGetRandomSatisfiedChannel(userGroup, modelRequest.Model)
+			channel, err = model.CacheGetRandomSatisfiedChannel(userGroup, modelRequest.Model, false)
 			if err != nil {
 				message := fmt.Sprintf("当前分组 %s 下对于模型 %s 无可用渠道", userGroup, modelRequest.Model)
 				if channel != nil {

middleware/recover.go

@@ -3,6 +3,7 @@ package middleware
 import (
 	"fmt"
 	"github.com/gin-gonic/gin"
+	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/common/logger"
 	"net/http"
 	"runtime/debug"
@@ -12,11 +13,15 @@ func RelayPanicRecover() gin.HandlerFunc {
 	return func(c *gin.Context) {
 		defer func() {
 			if err := recover(); err != nil {
-				logger.SysError(fmt.Sprintf("panic detected: %v", err))
+				ctx := c.Request.Context()
-				logger.SysError(fmt.Sprintf("stacktrace from panic: %s", string(debug.Stack())))
+				logger.Errorf(ctx, fmt.Sprintf("panic detected: %v", err))
+				logger.Errorf(ctx, fmt.Sprintf("stacktrace from panic: %s", string(debug.Stack())))
+				logger.Errorf(ctx, fmt.Sprintf("request: %s %s", c.Request.Method, c.Request.URL.Path))
+				body, _ := common.GetRequestBody(c)
+				logger.Errorf(ctx, fmt.Sprintf("request body: %s", string(body)))
 				c.JSON(http.StatusInternalServerError, gin.H{
 					"error": gin.H{
-						"message": fmt.Sprintf("Panic detected, error: %v. Please submit a issue here: https://github.com/songquanpeng/one-api", err),
+						"message": fmt.Sprintf("Panic detected, error: %v. Please submit an issue with the related log here: https://github.com/songquanpeng/one-api", err),
 						"type":    "one_api_panic",
 					},
 				})

model/cache.go

@@ -191,7 +191,7 @@ func SyncChannelCache(frequency int) {
 	}
 }
 
-func CacheGetRandomSatisfiedChannel(group string, model string) (*Channel, error) {
+func CacheGetRandomSatisfiedChannel(group string, model string, ignoreFirstPriority bool) (*Channel, error) {
 	if !config.MemoryCacheEnabled {
 		return GetRandomSatisfiedChannel(group, model)
 	}
@@ -213,5 +213,10 @@ func CacheGetRandomSatisfiedChannel(group string, model string) (*Channel, error
 		}
 	}
 	idx := rand.Intn(endIdx)
+	if ignoreFirstPriority {
+		if endIdx < len(channels) { // which means there are more than one priority
+			idx = common.RandRange(endIdx, len(channels))
+		}
+	}
 	return channels[idx], nil
 }

model/channel.go

@@ -13,7 +13,7 @@ import (
 type Channel struct {
 	Id                 int     `json:"id"`
 	Type               int     `json:"type" gorm:"default:0"`
-	Key                string  `json:"key" gorm:"not null;index"`
+	Key                string  `json:"key" gorm:"type:text"`
 	Status             int     `json:"status" gorm:"default:1"`
 	Name               string  `json:"name" gorm:"index"`
 	Weight             *uint   `json:"weight" gorm:"default:0"`
@@ -32,23 +32,22 @@ type Channel struct {
 	Config             string  `json:"config"`
 }
 
-func GetAllChannels(startIdx int, num int, selectAll bool) ([]*Channel, error) {
+func GetAllChannels(startIdx int, num int, scope string) ([]*Channel, error) {
 	var channels []*Channel
 	var err error
-	if selectAll {
+	switch scope {
+	case "all":
 		err = DB.Order("id desc").Find(&channels).Error
-	} else {
+	case "disabled":
+		err = DB.Order("id desc").Where("status = ? or status = ?", common.ChannelStatusAutoDisabled, common.ChannelStatusManuallyDisabled).Find(&channels).Error
+	default:
 		err = DB.Order("id desc").Limit(num).Offset(startIdx).Omit("key").Find(&channels).Error
 	}
 	return channels, err
 }
 
 func SearchChannels(keyword string) (channels []*Channel, err error) {
-	keyCol := "`key`"
+	err = DB.Omit("key").Where("id = ? or name LIKE ?", helper.String2Int(keyword), keyword+"%").Find(&channels).Error
-	if common.UsingPostgreSQL {
-		keyCol = `"key"`
-	}
-	err = DB.Omit("key").Where("id = ? or name LIKE ? or "+keyCol+" = ?", helper.String2Int(keyword), keyword+"%", keyword).Find(&channels).Error
 	return channels, err
 }
 

model/main.go

@@ -72,7 +72,7 @@ func chooseDB() (*gorm.DB, error) {
 func InitDB() (err error) {
 	db, err := chooseDB()
 	if err == nil {
-		if config.DebugEnabled {
+		if config.DebugSQLEnabled {
 			db = db.Debug()
 		}
 		DB = db

model/option.go

@@ -57,6 +57,8 @@ func InitOptionMap() {
 	config.OptionMap["WeChatServerAddress"] = ""
 	config.OptionMap["WeChatServerToken"] = ""
 	config.OptionMap["WeChatAccountQRCodeImageURL"] = ""
+	config.OptionMap["MessagePusherAddress"] = ""
+	config.OptionMap["MessagePusherToken"] = ""
 	config.OptionMap["TurnstileSiteKey"] = ""
 	config.OptionMap["TurnstileSecretKey"] = ""
 	config.OptionMap["QuotaForNewUser"] = strconv.Itoa(config.QuotaForNewUser)
@@ -79,6 +81,9 @@ func InitOptionMap() {
 func loadOptionsFromDatabase() {
 	options, _ := AllOption()
 	for _, option := range options {
+		if option.Key == "ModelRatio" {
+			option.Value = common.AddNewMissingRatio(option.Value)
+		}
 		err := updateOptionMap(option.Key, option.Value)
 		if err != nil {
 			logger.SysError("failed to update option map: " + err.Error())
@@ -179,6 +184,10 @@ func updateOptionMap(key string, value string) (err error) {
 		config.WeChatServerToken = value
 	case "WeChatAccountQRCodeImageURL":
 		config.WeChatAccountQRCodeImageURL = value
+	case "MessagePusherAddress":
+		config.MessagePusherAddress = value
+	case "MessagePusherToken":
+		config.MessagePusherToken = value
 	case "TurnstileSiteKey":
 		config.TurnstileSiteKey = value
 	case "TurnstileSecretKey":

model/token.go

@@ -7,6 +7,7 @@ import (
 	"github.com/songquanpeng/one-api/common/config"
 	"github.com/songquanpeng/one-api/common/helper"
 	"github.com/songquanpeng/one-api/common/logger"
+	"github.com/songquanpeng/one-api/common/message"
 	"gorm.io/gorm"
 )
 
@@ -213,7 +214,7 @@ func PreConsumeTokenQuota(tokenId int, quota int) (err error) {
 			}
 			if email != "" {
 				topUpLink := fmt.Sprintf("%s/topup", config.ServerAddress)
-				err = common.SendEmail(prompt, email,
+				err = message.SendEmail(prompt, email,
 					fmt.Sprintf("%s，当前剩余额度为 %d，为了不影响您的使用，请及时充值。<br/>充值链接：<a href='%s'>%s</a>", prompt, userQuota, topUpLink, topUpLink))
 				if err != nil {
 					logger.SysError("failed to send email" + err.Error())

model/user.go

@@ -4,6 +4,7 @@ import (
 	"errors"
 	"fmt"
 	"github.com/songquanpeng/one-api/common"
+	"github.com/songquanpeng/one-api/common/blacklist"
 	"github.com/songquanpeng/one-api/common/config"
 	"github.com/songquanpeng/one-api/common/helper"
 	"github.com/songquanpeng/one-api/common/logger"
@@ -40,7 +41,7 @@ func GetMaxUserId() int {
 }
 
 func GetAllUsers(startIdx int, num int) (users []*User, err error) {
-	err = DB.Order("id desc").Limit(num).Offset(startIdx).Omit("password").Find(&users).Error
+	err = DB.Order("id desc").Limit(num).Offset(startIdx).Omit("password").Where("status != ?", common.UserStatusDeleted).Find(&users).Error
 	return users, err
 }
 
@@ -123,6 +124,11 @@ func (user *User) Update(updatePassword bool) error {
 			return err
 		}
 	}
+	if user.Status == common.UserStatusDisabled {
+		blacklist.BanUser(user.Id)
+	} else if user.Status == common.UserStatusEnabled {
+		blacklist.UnbanUser(user.Id)
+	}
 	err = DB.Model(user).Updates(user).Error
 	return err
 }
@@ -131,7 +137,10 @@ func (user *User) Delete() error {
 	if user.Id == 0 {
 		return errors.New("id 为空！")
 	}
-	err := DB.Delete(user).Error
+	blacklist.BanUser(user.Id)
+	user.Username = fmt.Sprintf("deleted_%s", helper.GetUUID())
+	user.Status = common.UserStatusDeleted
+	err := DB.Model(user).Updates(user).Error
 	return err
 }
 

monitor/channel.go

@@ -0,0 +1,55 @@
+package monitor
+
+import (
+	"fmt"
+	"github.com/songquanpeng/one-api/common"
+	"github.com/songquanpeng/one-api/common/config"
+	"github.com/songquanpeng/one-api/common/logger"
+	"github.com/songquanpeng/one-api/common/message"
+	"github.com/songquanpeng/one-api/model"
+)
+
+func notifyRootUser(subject string, content string) {
+	if config.MessagePusherAddress != "" {
+		err := message.SendMessage(subject, content, content)
+		if err != nil {
+			logger.SysError(fmt.Sprintf("failed to send message: %s", err.Error()))
+		} else {
+			return
+		}
+	}
+	if config.RootUserEmail == "" {
+		config.RootUserEmail = model.GetRootUserEmail()
+	}
+	err := message.SendEmail(subject, config.RootUserEmail, content)
+	if err != nil {
+		logger.SysError(fmt.Sprintf("failed to send email: %s", err.Error()))
+	}
+}
+
+// DisableChannel disable & notify
+func DisableChannel(channelId int, channelName string, reason string) {
+	model.UpdateChannelStatusById(channelId, common.ChannelStatusAutoDisabled)
+	logger.SysLog(fmt.Sprintf("channel #%d has been disabled: %s", channelId, reason))
+	subject := fmt.Sprintf("通道「%s」（#%d）已被禁用", channelName, channelId)
+	content := fmt.Sprintf("通道「%s」（#%d）已被禁用，原因：%s", channelName, channelId, reason)
+	notifyRootUser(subject, content)
+}
+
+func MetricDisableChannel(channelId int, successRate float64) {
+	model.UpdateChannelStatusById(channelId, common.ChannelStatusAutoDisabled)
+	logger.SysLog(fmt.Sprintf("channel #%d has been disabled due to low success rate: %.2f", channelId, successRate*100))
+	subject := fmt.Sprintf("通道 #%d 已被禁用", channelId)
+	content := fmt.Sprintf("该渠道在最近 %d 次调用中成功率为 %.2f%%，低于阈值 %.2f%%，因此被系统自动禁用。",
+		config.MetricQueueSize, successRate*100, config.MetricSuccessRateThreshold*100)
+	notifyRootUser(subject, content)
+}
+
+// EnableChannel enable & notify
+func EnableChannel(channelId int, channelName string) {
+	model.UpdateChannelStatusById(channelId, common.ChannelStatusEnabled)
+	logger.SysLog(fmt.Sprintf("channel #%d has been enabled", channelId))
+	subject := fmt.Sprintf("通道「%s」（#%d）已被启用", channelName, channelId)
+	content := fmt.Sprintf("通道「%s」（#%d）已被启用", channelName, channelId)
+	notifyRootUser(subject, content)
+}

monitor/metric.go

@@ -0,0 +1,79 @@
+package monitor
+
+import (
+	"github.com/songquanpeng/one-api/common/config"
+)
+
+var store = make(map[int][]bool)
+var metricSuccessChan = make(chan int, config.MetricSuccessChanSize)
+var metricFailChan = make(chan int, config.MetricFailChanSize)
+
+func consumeSuccess(channelId int) {
+	if len(store[channelId]) > config.MetricQueueSize {
+		store[channelId] = store[channelId][1:]
+	}
+	store[channelId] = append(store[channelId], true)
+}
+
+func consumeFail(channelId int) (bool, float64) {
+	if len(store[channelId]) > config.MetricQueueSize {
+		store[channelId] = store[channelId][1:]
+	}
+	store[channelId] = append(store[channelId], false)
+	successCount := 0
+	for _, success := range store[channelId] {
+		if success {
+			successCount++
+		}
+	}
+	successRate := float64(successCount) / float64(len(store[channelId]))
+	if len(store[channelId]) < config.MetricQueueSize {
+		return false, successRate
+	}
+	if successRate < config.MetricSuccessRateThreshold {
+		store[channelId] = make([]bool, 0)
+		return true, successRate
+	}
+	return false, successRate
+}
+
+func metricSuccessConsumer() {
+	for {
+		select {
+		case channelId := <-metricSuccessChan:
+			consumeSuccess(channelId)
+		}
+	}
+}
+
+func metricFailConsumer() {
+	for {
+		select {
+		case channelId := <-metricFailChan:
+			disable, successRate := consumeFail(channelId)
+			if disable {
+				go MetricDisableChannel(channelId, successRate)
+			}
+		}
+	}
+}
+
+func init() {
+	if config.EnableMetric {
+		go metricSuccessConsumer()
+		go metricFailConsumer()
+	}
+}
+
+func Emit(channelId int, success bool) {
+	if !config.EnableMetric {
+		return
+	}
+	go func() {
+		if success {
+			metricSuccessChan <- channelId
+		} else {
+			metricFailChan <- channelId
+		}
+	}()
+}

relay/channel/ali/main.go

@@ -33,6 +33,9 @@ func ConvertRequest(request model.GeneralOpenAIRequest) *ChatRequest {
 		enableSearch = true
 		aliModel = strings.TrimSuffix(aliModel, EnableSearchModelSuffix)
 	}
+	if request.TopP >= 1 {
+		request.TopP = 0.9999
+	}
 	return &ChatRequest{
 		Model: aliModel,
 		Input: Input{
@@ -42,6 +45,9 @@ func ConvertRequest(request model.GeneralOpenAIRequest) *ChatRequest {
 			EnableSearch:      enableSearch,
 			IncrementalOutput: request.Stream,
 			Seed:              uint64(request.Seed),
+			MaxTokens:         request.MaxTokens,
+			Temperature:       request.Temperature,
+			TopP:              request.TopP,
 		},
 	}
 }

relay/channel/ali/model.go

@@ -16,6 +16,8 @@ type Parameters struct {
 	Seed              uint64  `json:"seed,omitempty"`
 	EnableSearch      bool    `json:"enable_search,omitempty"`
 	IncrementalOutput bool    `json:"incremental_output,omitempty"`
+	MaxTokens         int     `json:"max_tokens,omitempty"`
+	Temperature       float64 `json:"temperature,omitempty"`
 }
 
 type ChatRequest struct {

relay/channel/anthropic/adaptor.go

@@ -5,7 +5,6 @@ import (
 	"fmt"
 	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/relay/channel"
-	"github.com/songquanpeng/one-api/relay/channel/openai"
 	"github.com/songquanpeng/one-api/relay/model"
 	"github.com/songquanpeng/one-api/relay/util"
 	"io"
@@ -20,7 +19,7 @@ func (a *Adaptor) Init(meta *util.RelayMeta) {
 }
 
 func (a *Adaptor) GetRequestURL(meta *util.RelayMeta) (string, error) {
-	return fmt.Sprintf("%s/v1/complete", meta.BaseURL), nil
+	return fmt.Sprintf("%s/v1/messages", meta.BaseURL), nil
 }
 
 func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, meta *util.RelayMeta) error {
@@ -31,6 +30,7 @@ func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Request, meta *ut
 		anthropicVersion = "2023-06-01"
 	}
 	req.Header.Set("anthropic-version", anthropicVersion)
+	req.Header.Set("anthropic-beta", "messages-2023-12-15")
 	return nil
 }
 
@@ -47,9 +47,7 @@ func (a *Adaptor) DoRequest(c *gin.Context, meta *util.RelayMeta, requestBody io
 
 func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, meta *util.RelayMeta) (usage *model.Usage, err *model.ErrorWithStatusCode) {
 	if meta.IsStream {
-		var responseText string
+		err, usage = StreamHandler(c, resp)
-		err, responseText = StreamHandler(c, resp)
-		usage = openai.ResponseText2Usage(responseText, meta.ActualModelName, meta.PromptTokens)
 	} else {
 		err, usage = Handler(c, resp, meta.PromptTokens, meta.ActualModelName)
 	}

relay/channel/anthropic/constants.go

@@ -1,5 +1,8 @@
 package anthropic
 
 var ModelList = []string{
-	"claude-instant-1", "claude-2", "claude-2.0", "claude-2.1",
+	"claude-instant-1.2", "claude-2.0", "claude-2.1",
+	"claude-3-haiku-20240229",
+	"claude-3-sonnet-20240229",
+	"claude-3-opus-20240229",
 }

relay/channel/anthropic/main.go

@@ -7,6 +7,7 @@ import (
 	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/common/helper"
+	"github.com/songquanpeng/one-api/common/image"
 	"github.com/songquanpeng/one-api/common/logger"
 	"github.com/songquanpeng/one-api/relay/channel/openai"
 	"github.com/songquanpeng/one-api/relay/model"
@@ -15,73 +16,135 @@ import (
 	"strings"
 )
 
-func stopReasonClaude2OpenAI(reason string) string {
+func stopReasonClaude2OpenAI(reason *string) string {
-	switch reason {
+	if reason == nil {
+		return ""
+	}
+	switch *reason {
+	case "end_turn":
+		return "stop"
 	case "stop_sequence":
 		return "stop"
 	case "max_tokens":
 		return "length"
 	default:
-		return reason
+		return *reason
 	}
 }
 
 func ConvertRequest(textRequest model.GeneralOpenAIRequest) *Request {
 	claudeRequest := Request{
-		Model:             textRequest.Model,
+		Model:       textRequest.Model,
-		Prompt:            "",
+		MaxTokens:   textRequest.MaxTokens,
-		MaxTokensToSample: textRequest.MaxTokens,
+		Temperature: textRequest.Temperature,
-		StopSequences:     nil,
+		TopP:        textRequest.TopP,
-		Temperature:       textRequest.Temperature,
+		Stream:      textRequest.Stream,
-		TopP:              textRequest.TopP,
-		Stream:            textRequest.Stream,
 	}
-	if claudeRequest.MaxTokensToSample == 0 {
+	if claudeRequest.MaxTokens == 0 {
-		claudeRequest.MaxTokensToSample = 1000000
+		claudeRequest.MaxTokens = 4096
+	}
+	// legacy model name mapping
+	if claudeRequest.Model == "claude-instant-1" {
+		claudeRequest.Model = "claude-instant-1.1"
+	} else if claudeRequest.Model == "claude-2" {
+		claudeRequest.Model = "claude-2.1"
 	}
-	prompt := ""
 	for _, message := range textRequest.Messages {
-		if message.Role == "user" {
+		if message.Role == "system" && claudeRequest.System == "" {
-			prompt += fmt.Sprintf("\n\nHuman: %s", message.Content)
+			claudeRequest.System = message.StringContent()
-		} else if message.Role == "assistant" {
+			continue
-			prompt += fmt.Sprintf("\n\nAssistant: %s", message.Content)
-		} else if message.Role == "system" {
-			if prompt == "" {
-				prompt = message.StringContent()
-			}
 		}
+		claudeMessage := Message{
+			Role: message.Role,
+		}
+		var content Content
+		if message.IsStringContent() {
+			content.Type = "text"
+			content.Text = message.StringContent()
+			claudeMessage.Content = append(claudeMessage.Content, content)
+			claudeRequest.Messages = append(claudeRequest.Messages, claudeMessage)
+			continue
+		}
+		var contents []Content
+		openaiContent := message.ParseContent()
+		for _, part := range openaiContent {
+			var content Content
+			if part.Type == model.ContentTypeText {
+				content.Type = "text"
+				content.Text = part.Text
+			} else if part.Type == model.ContentTypeImageURL {
+				content.Type = "image"
+				content.Source = &ImageSource{
+					Type: "base64",
+				}
+				mimeType, data, _ := image.GetImageFromUrl(part.ImageURL.Url)
+				content.Source.MediaType = mimeType
+				content.Source.Data = data
+			}
+			contents = append(contents, content)
+		}
+		claudeMessage.Content = contents
+		claudeRequest.Messages = append(claudeRequest.Messages, claudeMessage)
 	}
-	prompt += "\n\nAssistant:"
-	claudeRequest.Prompt = prompt
 	return &claudeRequest
 }
 
-func streamResponseClaude2OpenAI(claudeResponse *Response) *openai.ChatCompletionsStreamResponse {
+// https://docs.anthropic.com/claude/reference/messages-streaming
+func streamResponseClaude2OpenAI(claudeResponse *StreamResponse) (*openai.ChatCompletionsStreamResponse, *Response) {
+	var response *Response
+	var responseText string
+	var stopReason string
+	switch claudeResponse.Type {
+	case "message_start":
+		return nil, claudeResponse.Message
+	case "content_block_start":
+		if claudeResponse.ContentBlock != nil {
+			responseText = claudeResponse.ContentBlock.Text
+		}
+	case "content_block_delta":
+		if claudeResponse.Delta != nil {
+			responseText = claudeResponse.Delta.Text
+		}
+	case "message_delta":
+		if claudeResponse.Usage != nil {
+			response = &Response{
+				Usage: *claudeResponse.Usage,
+			}
+		}
+		if claudeResponse.Delta != nil && claudeResponse.Delta.StopReason != nil {
+			stopReason = *claudeResponse.Delta.StopReason
+		}
+	}
 	var choice openai.ChatCompletionsStreamResponseChoice
-	choice.Delta.Content = claudeResponse.Completion
+	choice.Delta.Content = responseText
-	finishReason := stopReasonClaude2OpenAI(claudeResponse.StopReason)
+	choice.Delta.Role = "assistant"
+	finishReason := stopReasonClaude2OpenAI(&stopReason)
 	if finishReason != "null" {
 		choice.FinishReason = &finishReason
 	}
-	var response openai.ChatCompletionsStreamResponse
+	var openaiResponse openai.ChatCompletionsStreamResponse
-	response.Object = "chat.completion.chunk"
+	openaiResponse.Object = "chat.completion.chunk"
-	response.Model = claudeResponse.Model
+	openaiResponse.Choices = []openai.ChatCompletionsStreamResponseChoice{choice}
-	response.Choices = []openai.ChatCompletionsStreamResponseChoice{choice}
+	return &openaiResponse, response
-	return &response
 }
 
 func responseClaude2OpenAI(claudeResponse *Response) *openai.TextResponse {
+	var responseText string
+	if len(claudeResponse.Content) > 0 {
+		responseText = claudeResponse.Content[0].Text
+	}
 	choice := openai.TextResponseChoice{
 		Index: 0,
 		Message: model.Message{
 			Role:    "assistant",
-			Content: strings.TrimPrefix(claudeResponse.Completion, " "),
+			Content: responseText,
 			Name:    nil,
 		},
 		FinishReason: stopReasonClaude2OpenAI(claudeResponse.StopReason),
 	}
 	fullTextResponse := openai.TextResponse{
-		Id:      fmt.Sprintf("chatcmpl-%s", helper.GetUUID()),
+		Id:      fmt.Sprintf("chatcmpl-%s", claudeResponse.Id),
+		Model:   claudeResponse.Model,
 		Object:  "chat.completion",
 		Created: helper.GetTimestamp(),
 		Choices: []openai.TextResponseChoice{choice},
@@ -89,17 +152,15 @@ func responseClaude2OpenAI(claudeResponse *Response) *openai.TextResponse {
 	return &fullTextResponse
 }
 
-func StreamHandler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusCode, string) {
+func StreamHandler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusCode, *model.Usage) {
-	responseText := ""
-	responseId := fmt.Sprintf("chatcmpl-%s", helper.GetUUID())
 	createdTime := helper.GetTimestamp()
 	scanner := bufio.NewScanner(resp.Body)
 	scanner.Split(func(data []byte, atEOF bool) (advance int, token []byte, err error) {
 		if atEOF && len(data) == 0 {
 			return 0, nil, nil
 		}
-		if i := strings.Index(string(data), "\r\n\r\n"); i >= 0 {
+		if i := strings.Index(string(data), "\n"); i >= 0 {
-			return i + 4, data[0:i], nil
+			return i + 1, data[0:i], nil
 		}
 		if atEOF {
 			return len(data), data, nil
@@ -111,29 +172,45 @@ func StreamHandler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusC
 	go func() {
 		for scanner.Scan() {
 			data := scanner.Text()
-			if !strings.HasPrefix(data, "event: completion") {
+			if len(data) < 6 {
 				continue
 			}
-			data = strings.TrimPrefix(data, "event: completion\r\ndata: ")
+			if !strings.HasPrefix(data, "data: ") {
+				continue
+			}
+			data = strings.TrimPrefix(data, "data: ")
 			dataChan <- data
 		}
 		stopChan <- true
 	}()
 	common.SetEventStreamHeaders(c)
+	var usage model.Usage
+	var modelName string
+	var id string
 	c.Stream(func(w io.Writer) bool {
 		select {
 		case data := <-dataChan:
 			// some implementations may add \r at the end of data
 			data = strings.TrimSuffix(data, "\r")
-			var claudeResponse Response
+			var claudeResponse StreamResponse
 			err := json.Unmarshal([]byte(data), &claudeResponse)
 			if err != nil {
 				logger.SysError("error unmarshalling stream response: " + err.Error())
 				return true
 			}
-			responseText += claudeResponse.Completion
+			response, meta := streamResponseClaude2OpenAI(&claudeResponse)
-			response := streamResponseClaude2OpenAI(&claudeResponse)
+			if meta != nil {
-			response.Id = responseId
+				usage.PromptTokens += meta.Usage.InputTokens
+				usage.CompletionTokens += meta.Usage.OutputTokens
+				modelName = meta.Model
+				id = fmt.Sprintf("chatcmpl-%s", meta.Id)
+				return true
+			}
+			if response == nil {
+				return true
+			}
+			response.Id = id
+			response.Model = modelName
 			response.Created = createdTime
 			jsonStr, err := json.Marshal(response)
 			if err != nil {
@@ -147,11 +224,8 @@ func StreamHandler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusC
 			return false
 		}
 	})
-	err := resp.Body.Close()
+	_ = resp.Body.Close()
-	if err != nil {
+	return nil, &usage
-		return openai.ErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), ""
-	}
-	return nil, responseText
 }
 
 func Handler(c *gin.Context, resp *http.Response, promptTokens int, modelName string) (*model.ErrorWithStatusCode, *model.Usage) {
@@ -181,11 +255,10 @@ func Handler(c *gin.Context, resp *http.Response, promptTokens int, modelName st
 	}
 	fullTextResponse := responseClaude2OpenAI(&claudeResponse)
 	fullTextResponse.Model = modelName
-	completionTokens := openai.CountTokenText(claudeResponse.Completion, modelName)
 	usage := model.Usage{
-		PromptTokens:     promptTokens,
+		PromptTokens:     claudeResponse.Usage.InputTokens,
-		CompletionTokens: completionTokens,
+		CompletionTokens: claudeResponse.Usage.OutputTokens,
-		TotalTokens:      promptTokens + completionTokens,
+		TotalTokens:      claudeResponse.Usage.InputTokens + claudeResponse.Usage.OutputTokens,
 	}
 	fullTextResponse.Usage = usage
 	jsonResponse, err := json.Marshal(fullTextResponse)

relay/channel/anthropic/model.go

@@ -1,19 +1,44 @@
 package anthropic
 
+// https://docs.anthropic.com/claude/reference/messages_post
+
 type Metadata struct {
 	UserId string `json:"user_id"`
 }
 
+type ImageSource struct {
+	Type      string `json:"type"`
+	MediaType string `json:"media_type"`
+	Data      string `json:"data"`
+}
+
+type Content struct {
+	Type   string       `json:"type"`
+	Text   string       `json:"text,omitempty"`
+	Source *ImageSource `json:"source,omitempty"`
+}
+
+type Message struct {
+	Role    string    `json:"role"`
+	Content []Content `json:"content"`
+}
+
 type Request struct {
-	Model             string   `json:"model"`
+	Model         string    `json:"model"`
-	Prompt            string   `json:"prompt"`
+	Messages      []Message `json:"messages"`
-	MaxTokensToSample int      `json:"max_tokens_to_sample"`
+	System        string    `json:"system,omitempty"`
-	StopSequences     []string `json:"stop_sequences,omitempty"`
+	MaxTokens     int       `json:"max_tokens,omitempty"`
-	Temperature       float64  `json:"temperature,omitempty"`
+	StopSequences []string  `json:"stop_sequences,omitempty"`
-	TopP              float64  `json:"top_p,omitempty"`
+	Stream        bool      `json:"stream,omitempty"`
-	TopK              int      `json:"top_k,omitempty"`
+	Temperature   float64   `json:"temperature,omitempty"`
+	TopP          float64   `json:"top_p,omitempty"`
+	TopK          int       `json:"top_k,omitempty"`
 	//Metadata    `json:"metadata,omitempty"`
-	Stream bool `json:"stream,omitempty"`
+}
+
+type Usage struct {
+	InputTokens  int `json:"input_tokens"`
+	OutputTokens int `json:"output_tokens"`
 }
 
 type Error struct {
@@ -22,8 +47,29 @@ type Error struct {
 }
 
 type Response struct {
-	Completion string `json:"completion"`
+	Id           string    `json:"id"`
-	StopReason string `json:"stop_reason"`
+	Type         string    `json:"type"`
-	Model      string `json:"model"`
+	Role         string    `json:"role"`
-	Error      Error  `json:"error"`
+	Content      []Content `json:"content"`
+	Model        string    `json:"model"`
+	StopReason   *string   `json:"stop_reason"`
+	StopSequence *string   `json:"stop_sequence"`
+	Usage        Usage     `json:"usage"`
+	Error        Error     `json:"error"`
+}
+
+type Delta struct {
+	Type         string  `json:"type"`
+	Text         string  `json:"text"`
+	StopReason   *string `json:"stop_reason"`
+	StopSequence *string `json:"stop_sequence"`
+}
+
+type StreamResponse struct {
+	Type         string    `json:"type"`
+	Message      *Response `json:"message"`
+	Index        int       `json:"index"`
+	ContentBlock *Content  `json:"content_block"`
+	Delta        *Delta    `json:"delta"`
+	Usage        *Usage    `json:"usage"`
 }

relay/channel/baidu/adaptor.go

@@ -2,6 +2,7 @@ package baidu
 
 import (
 	"errors"
+	"fmt"
 	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/relay/channel"
 	"github.com/songquanpeng/one-api/relay/constant"
@@ -9,6 +10,7 @@ import (
 	"github.com/songquanpeng/one-api/relay/util"
 	"io"
 	"net/http"
+	"strings"
 )
 
 type Adaptor struct {
@@ -20,23 +22,33 @@ func (a *Adaptor) Init(meta *util.RelayMeta) {
 
 func (a *Adaptor) GetRequestURL(meta *util.RelayMeta) (string, error) {
 	// https://cloud.baidu.com/doc/WENXINWORKSHOP/s/clntwmv7t
-	var fullRequestURL string
+	suffix := "chat/"
-	switch meta.ActualModelName {
+	if strings.HasPrefix("Embedding", meta.ActualModelName) {
-	case "ERNIE-Bot-4":
+		suffix = "embeddings/"
-		fullRequestURL = "https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/completions_pro"
-	case "ERNIE-Bot-8K":
-		fullRequestURL = "https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/ernie_bot_8k"
-	case "ERNIE-Bot":
-		fullRequestURL = "https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/completions"
-	case "ERNIE-Speed":
-		fullRequestURL = "https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/ernie_speed"
-	case "ERNIE-Bot-turbo":
-		fullRequestURL = "https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/eb-instant"
-	case "BLOOMZ-7B":
-		fullRequestURL = "https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/bloomz_7b1"
-	case "Embedding-V1":
-		fullRequestURL = "https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/embeddings/embedding-v1"
 	}
+	switch meta.ActualModelName {
+	case "ERNIE-4.0":
+		suffix += "completions_pro"
+	case "ERNIE-Bot-4":
+		suffix += "completions_pro"
+	case "ERNIE-3.5-8K":
+		suffix += "completions"
+	case "ERNIE-Bot-8K":
+		suffix += "ernie_bot_8k"
+	case "ERNIE-Bot":
+		suffix += "completions"
+	case "ERNIE-Speed":
+		suffix += "ernie_speed"
+	case "ERNIE-Bot-turbo":
+		suffix += "eb-instant"
+	case "BLOOMZ-7B":
+		suffix += "bloomz_7b1"
+	case "Embedding-V1":
+		suffix += "embedding-v1"
+	default:
+		suffix += meta.ActualModelName
+	}
+	fullRequestURL := fmt.Sprintf("%s/rpc/2.0/ai_custom/v1/wenxinworkshop/%s", meta.BaseURL, suffix)
 	var accessToken string
 	var err error
 	if accessToken, err = GetAccessToken(meta.APIKey); err != nil {

relay/channel/gemini/constants.go

@@ -1,6 +1,6 @@
 package gemini
 
 var ModelList = []string{
-	"gemini-pro",
+	"gemini-pro", "gemini-1.0-pro-001",
-	"gemini-pro-vision",
+	"gemini-pro-vision", "gemini-1.0-pro-vision-001",
 }

relay/channel/groq/constants.go

@@ -0,0 +1,10 @@
+package groq
+
+// https://console.groq.com/docs/models
+
+var ModelList = []string{
+	"gemma-7b-it",
+	"llama2-7b-2048",
+	"llama2-70b-4096",
+	"mixtral-8x7b-32768",
+}

relay/channel/mistral/constants.go

@@ -0,0 +1,10 @@
+package mistral
+
+var ModelList = []string{
+	"open-mistral-7b",
+	"open-mixtral-8x7b",
+	"mistral-small-latest",
+	"mistral-medium-latest",
+	"mistral-large-latest",
+	"mistral-embed",
+}

relay/channel/openai/adaptor.go

@@ -6,10 +6,7 @@ import (
 	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/relay/channel"
-	"github.com/songquanpeng/one-api/relay/channel/ai360"
-	"github.com/songquanpeng/one-api/relay/channel/baichuan"
 	"github.com/songquanpeng/one-api/relay/channel/minimax"
-	"github.com/songquanpeng/one-api/relay/channel/moonshot"
 	"github.com/songquanpeng/one-api/relay/model"
 	"github.com/songquanpeng/one-api/relay/util"
 	"io"
@@ -85,33 +82,11 @@ func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, meta *util.Rel
 }
 
 func (a *Adaptor) GetModelList() []string {
-	switch a.ChannelType {
+	_, modelList := GetCompatibleChannelMeta(a.ChannelType)
-	case common.ChannelType360:
+	return modelList
-		return ai360.ModelList
-	case common.ChannelTypeMoonshot:
-		return moonshot.ModelList
-	case common.ChannelTypeBaichuan:
-		return baichuan.ModelList
-	case common.ChannelTypeMinimax:
-		return minimax.ModelList
-	default:
-		return ModelList
-	}
 }
 
 func (a *Adaptor) GetChannelName() string {
-	switch a.ChannelType {
+	channelName, _ := GetCompatibleChannelMeta(a.ChannelType)
-	case common.ChannelTypeAzure:
+	return channelName
-		return "azure"
-	case common.ChannelType360:
-		return "360"
-	case common.ChannelTypeMoonshot:
-		return "moonshot"
-	case common.ChannelTypeBaichuan:
-		return "baichuan"
-	case common.ChannelTypeMinimax:
-		return "minimax"
-	default:
-		return "openai"
-	}
 }

relay/channel/openai/compatible.go

@@ -0,0 +1,42 @@
+package openai
+
+import (
+	"github.com/songquanpeng/one-api/common"
+	"github.com/songquanpeng/one-api/relay/channel/ai360"
+	"github.com/songquanpeng/one-api/relay/channel/baichuan"
+	"github.com/songquanpeng/one-api/relay/channel/groq"
+	"github.com/songquanpeng/one-api/relay/channel/minimax"
+	"github.com/songquanpeng/one-api/relay/channel/mistral"
+	"github.com/songquanpeng/one-api/relay/channel/moonshot"
+)
+
+var CompatibleChannels = []int{
+	common.ChannelTypeAzure,
+	common.ChannelType360,
+	common.ChannelTypeMoonshot,
+	common.ChannelTypeBaichuan,
+	common.ChannelTypeMinimax,
+	common.ChannelTypeMistral,
+	common.ChannelTypeGroq,
+}
+
+func GetCompatibleChannelMeta(channelType int) (string, []string) {
+	switch channelType {
+	case common.ChannelTypeAzure:
+		return "azure", ModelList
+	case common.ChannelType360:
+		return "360", ai360.ModelList
+	case common.ChannelTypeMoonshot:
+		return "moonshot", moonshot.ModelList
+	case common.ChannelTypeBaichuan:
+		return "baichuan", baichuan.ModelList
+	case common.ChannelTypeMinimax:
+		return "minimax", minimax.ModelList
+	case common.ChannelTypeMistral:
+		return "mistralai", mistral.ModelList
+	case common.ChannelTypeGroq:
+		return "groq", groq.ModelList
+	default:
+		return "openai", ModelList
+	}
+}

relay/channel/tencent/main.go

@@ -28,17 +28,6 @@ func ConvertRequest(request model.GeneralOpenAIRequest) *ChatRequest {
 	messages := make([]Message, 0, len(request.Messages))
 	for i := 0; i < len(request.Messages); i++ {
 		message := request.Messages[i]
-		if message.Role == "system" {
-			messages = append(messages, Message{
-				Role:    "user",
-				Content: message.StringContent(),
-			})
-			messages = append(messages, Message{
-				Role:    "assistant",
-				Content: "Okay",
-			})
-			continue
-		}
 		messages = append(messages, Message{
 			Content: message.StringContent(),
 			Role:    message.Role,

relay/channel/xunfei/main.go

@@ -27,21 +27,10 @@ import (
 func requestOpenAI2Xunfei(request model.GeneralOpenAIRequest, xunfeiAppId string, domain string) *ChatRequest {
 	messages := make([]Message, 0, len(request.Messages))
 	for _, message := range request.Messages {
-		if message.Role == "system" {
+		messages = append(messages, Message{
-			messages = append(messages, Message{
+			Role:    message.Role,
-				Role:    "user",
+			Content: message.StringContent(),
-				Content: message.StringContent(),
+		})
-			})
-			messages = append(messages, Message{
-				Role:    "assistant",
-				Content: "Okay",
-			})
-		} else {
-			messages = append(messages, Message{
-				Role:    message.Role,
-				Content: message.StringContent(),
-			})
-		}
 	}
 	xunfeiRequest := ChatRequest{}
 	xunfeiRequest.Header.AppId = xunfeiAppId

relay/channel/zhipu/main.go

@@ -76,21 +76,10 @@ func GetToken(apikey string) string {
 func ConvertRequest(request model.GeneralOpenAIRequest) *Request {
 	messages := make([]Message, 0, len(request.Messages))
 	for _, message := range request.Messages {
-		if message.Role == "system" {
+		messages = append(messages, Message{
-			messages = append(messages, Message{
+			Role:    message.Role,
-				Role:    "system",
+			Content: message.StringContent(),
-				Content: message.StringContent(),
+		})
-			})
-			messages = append(messages, Message{
-				Role:    "user",
-				Content: "Okay",
-			})
-		} else {
-			messages = append(messages, Message{
-				Role:    message.Role,
-				Content: message.StringContent(),
-			})
-		}
 	}
 	return &Request{
 		Prompt:      messages,

relay/constant/image.go

@@ -0,0 +1,24 @@
+package constant
+
+var DalleSizeRatios = map[string]map[string]float64{
+	"dall-e-2": {
+		"256x256":   1,
+		"512x512":   1.125,
+		"1024x1024": 1.25,
+	},
+	"dall-e-3": {
+		"1024x1024": 1,
+		"1024x1792": 2,
+		"1792x1024": 2,
+	},
+}
+
+var DalleGenerationImageAmounts = map[string][2]int{
+	"dall-e-2": {1, 10},
+	"dall-e-3": {1, 1}, // OpenAI allows n=1 currently.
+}
+
+var DalleImagePromptLengthLimitations = map[string]int{
+	"dall-e-2": 1000,
+	"dall-e-3": 4000,
+}

relay/controller/helper.go

@@ -36,6 +36,65 @@ func getAndValidateTextRequest(c *gin.Context, relayMode int) (*relaymodel.Gener
 	return textRequest, nil
 }
 
+func getImageRequest(c *gin.Context, relayMode int) (*openai.ImageRequest, error) {
+	imageRequest := &openai.ImageRequest{}
+	err := common.UnmarshalBodyReusable(c, imageRequest)
+	if err != nil {
+		return nil, err
+	}
+	if imageRequest.N == 0 {
+		imageRequest.N = 1
+	}
+	if imageRequest.Size == "" {
+		imageRequest.Size = "1024x1024"
+	}
+	if imageRequest.Model == "" {
+		imageRequest.Model = "dall-e-2"
+	}
+	return imageRequest, nil
+}
+
+func validateImageRequest(imageRequest *openai.ImageRequest, meta *util.RelayMeta) *relaymodel.ErrorWithStatusCode {
+	// model validation
+	_, hasValidSize := constant.DalleSizeRatios[imageRequest.Model][imageRequest.Size]
+	if !hasValidSize {
+		return openai.ErrorWrapper(errors.New("size not supported for this image model"), "size_not_supported", http.StatusBadRequest)
+	}
+	// check prompt length
+	if imageRequest.Prompt == "" {
+		return openai.ErrorWrapper(errors.New("prompt is required"), "prompt_missing", http.StatusBadRequest)
+	}
+	if len(imageRequest.Prompt) > constant.DalleImagePromptLengthLimitations[imageRequest.Model] {
+		return openai.ErrorWrapper(errors.New("prompt is too long"), "prompt_too_long", http.StatusBadRequest)
+	}
+	// Number of generated images validation
+	if !isWithinRange(imageRequest.Model, imageRequest.N) {
+		// channel not azure
+		if meta.ChannelType != common.ChannelTypeAzure {
+			return openai.ErrorWrapper(errors.New("invalid value of n"), "n_not_within_range", http.StatusBadRequest)
+		}
+	}
+	return nil
+}
+
+func getImageCostRatio(imageRequest *openai.ImageRequest) (float64, error) {
+	if imageRequest == nil {
+		return 0, errors.New("imageRequest is nil")
+	}
+	imageCostRatio, hasValidSize := constant.DalleSizeRatios[imageRequest.Model][imageRequest.Size]
+	if !hasValidSize {
+		return 0, fmt.Errorf("size not supported for this image model: %s", imageRequest.Size)
+	}
+	if imageRequest.Quality == "hd" && imageRequest.Model == "dall-e-3" {
+		if imageRequest.Size == "1024x1024" {
+			imageCostRatio *= 2
+		} else {
+			imageCostRatio *= 1.5
+		}
+	}
+	return imageCostRatio, nil
+}
+
 func getPromptTokens(textRequest *relaymodel.GeneralOpenAIRequest, relayMode int) int {
 	switch relayMode {
 	case constant.RelayModeChatCompletions:
@@ -113,10 +172,8 @@ func postConsumeQuota(ctx context.Context, usage *relaymodel.Usage, meta *util.R
 	if err != nil {
 		logger.Error(ctx, "error update user quota cache: "+err.Error())
 	}
-	if quota != 0 {
+	logContent := fmt.Sprintf("模型倍率 %.2f，分组倍率 %.2f，补全倍率 %.2f", modelRatio, groupRatio, completionRatio)
-		logContent := fmt.Sprintf("模型倍率 %.2f，分组倍率 %.2f，补全倍率 %.2f", modelRatio, groupRatio, completionRatio)
+	model.RecordConsumeLog(ctx, meta.UserId, meta.ChannelId, promptTokens, completionTokens, textRequest.Model, meta.TokenName, quota, logContent)
-		model.RecordConsumeLog(ctx, meta.UserId, meta.ChannelId, promptTokens, completionTokens, textRequest.Model, meta.TokenName, quota, logContent)
+	model.UpdateUserUsedQuotaAndRequestCount(meta.UserId, quota)
-		model.UpdateUserUsedQuotaAndRequestCount(meta.UserId, quota)
+	model.UpdateChannelUsedQuota(meta.ChannelId, quota)
-		model.UpdateChannelUsedQuota(meta.ChannelId, quota)
-	}
 }

relay/controller/image.go

@@ -10,6 +10,7 @@ import (
 	"github.com/songquanpeng/one-api/common/logger"
 	"github.com/songquanpeng/one-api/model"
 	"github.com/songquanpeng/one-api/relay/channel/openai"
+	"github.com/songquanpeng/one-api/relay/constant"
 	relaymodel "github.com/songquanpeng/one-api/relay/model"
 	"github.com/songquanpeng/one-api/relay/util"
 	"io"
@@ -20,120 +21,65 @@ import (
 )
 
 func isWithinRange(element string, value int) bool {
-	if _, ok := common.DalleGenerationImageAmounts[element]; !ok {
+	if _, ok := constant.DalleGenerationImageAmounts[element]; !ok {
 		return false
 	}
-	min := common.DalleGenerationImageAmounts[element][0]
+	min := constant.DalleGenerationImageAmounts[element][0]
-	max := common.DalleGenerationImageAmounts[element][1]
+	max := constant.DalleGenerationImageAmounts[element][1]
 
 	return value >= min && value <= max
 }
 
 func RelayImageHelper(c *gin.Context, relayMode int) *relaymodel.ErrorWithStatusCode {
-	imageModel := "dall-e-2"
+	ctx := c.Request.Context()
-	imageSize := "1024x1024"
+	meta := util.GetRelayMeta(c)
-
+	imageRequest, err := getImageRequest(c, meta.Mode)
-	tokenId := c.GetInt("token_id")
-	channelType := c.GetInt("channel")
-	channelId := c.GetInt("channel_id")
-	userId := c.GetInt("id")
-	group := c.GetString("group")
-
-	var imageRequest openai.ImageRequest
-	err := common.UnmarshalBodyReusable(c, &imageRequest)
 	if err != nil {
-		return openai.ErrorWrapper(err, "bind_request_body_failed", http.StatusBadRequest)
+		logger.Errorf(ctx, "getImageRequest failed: %s", err.Error())
-	}
+		return openai.ErrorWrapper(err, "invalid_image_request", http.StatusBadRequest)
-
-	if imageRequest.N == 0 {
-		imageRequest.N = 1
-	}
-
-	// Size validation
-	if imageRequest.Size != "" {
-		imageSize = imageRequest.Size
-	}
-
-	// Model validation
-	if imageRequest.Model != "" {
-		imageModel = imageRequest.Model
-	}
-
-	imageCostRatio, hasValidSize := common.DalleSizeRatios[imageModel][imageSize]
-
-	// Check if model is supported
-	if hasValidSize {
-		if imageRequest.Quality == "hd" && imageModel == "dall-e-3" {
-			if imageSize == "1024x1024" {
-				imageCostRatio *= 2
-			} else {
-				imageCostRatio *= 1.5
-			}
-		}
-	} else {
-		return openai.ErrorWrapper(errors.New("size not supported for this image model"), "size_not_supported", http.StatusBadRequest)
-	}
-
-	// Prompt validation
-	if imageRequest.Prompt == "" {
-		return openai.ErrorWrapper(errors.New("prompt is required"), "prompt_missing", http.StatusBadRequest)
-	}
-
-	// Check prompt length
-	if len(imageRequest.Prompt) > common.DalleImagePromptLengthLimitations[imageModel] {
-		return openai.ErrorWrapper(errors.New("prompt is too long"), "prompt_too_long", http.StatusBadRequest)
-	}
-
-	// Number of generated images validation
-	if !isWithinRange(imageModel, imageRequest.N) {
-		// channel not azure
-		if channelType != common.ChannelTypeAzure {
-			return openai.ErrorWrapper(errors.New("invalid value of n"), "n_not_within_range", http.StatusBadRequest)
-		}
 	}
 
 	// map model name
-	modelMapping := c.GetString("model_mapping")
+	var isModelMapped bool
-	isModelMapped := false
+	meta.OriginModelName = imageRequest.Model
-	if modelMapping != "" {
+	imageRequest.Model, isModelMapped = util.GetMappedModelName(imageRequest.Model, meta.ModelMapping)
-		modelMap := make(map[string]string)
+	meta.ActualModelName = imageRequest.Model
-		err := json.Unmarshal([]byte(modelMapping), &modelMap)
+
-		if err != nil {
+	// model validation
-			return openai.ErrorWrapper(err, "unmarshal_model_mapping_failed", http.StatusInternalServerError)
+	bizErr := validateImageRequest(imageRequest, meta)
-		}
+	if bizErr != nil {
-		if modelMap[imageModel] != "" {
+		return bizErr
-			imageModel = modelMap[imageModel]
-			isModelMapped = true
-		}
 	}
-	baseURL := common.ChannelBaseURLs[channelType]
+
+	imageCostRatio, err := getImageCostRatio(imageRequest)
+	if err != nil {
+		return openai.ErrorWrapper(err, "get_image_cost_ratio_failed", http.StatusInternalServerError)
+	}
+
 	requestURL := c.Request.URL.String()
-	if c.GetString("base_url") != "" {
+	fullRequestURL := util.GetFullRequestURL(meta.BaseURL, requestURL, meta.ChannelType)
-		baseURL = c.GetString("base_url")
+	if meta.ChannelType == common.ChannelTypeAzure {
-	}
-	fullRequestURL := util.GetFullRequestURL(baseURL, requestURL, channelType)
-	if channelType == common.ChannelTypeAzure {
 		// https://learn.microsoft.com/en-us/azure/ai-services/openai/dall-e-quickstart?tabs=dalle3%2Ccommand-line&pivots=rest-api
 		apiVersion := util.GetAzureAPIVersion(c)
 		// https://{resource_name}.openai.azure.com/openai/deployments/dall-e-3/images/generations?api-version=2023-06-01-preview
-		fullRequestURL = fmt.Sprintf("%s/openai/deployments/%s/images/generations?api-version=%s", baseURL, imageModel, apiVersion)
+		fullRequestURL = fmt.Sprintf("%s/openai/deployments/%s/images/generations?api-version=%s", meta.BaseURL, imageRequest.Model, apiVersion)
 	}
 
 	var requestBody io.Reader
-	if isModelMapped || channelType == common.ChannelTypeAzure { // make Azure channel request body
+	if isModelMapped || meta.ChannelType == common.ChannelTypeAzure { // make Azure channel request body
 		jsonStr, err := json.Marshal(imageRequest)
 		if err != nil {
-			return openai.ErrorWrapper(err, "marshal_text_request_failed", http.StatusInternalServerError)
+			return openai.ErrorWrapper(err, "marshal_image_request_failed", http.StatusInternalServerError)
 		}
 		requestBody = bytes.NewBuffer(jsonStr)
 	} else {
 		requestBody = c.Request.Body
 	}
 
-	modelRatio := common.GetModelRatio(imageModel)
+	modelRatio := common.GetModelRatio(imageRequest.Model)
-	groupRatio := common.GetGroupRatio(group)
+	groupRatio := common.GetGroupRatio(meta.Group)
 	ratio := modelRatio * groupRatio
-	userQuota, err := model.CacheGetUserQuota(userId)
+	userQuota, err := model.CacheGetUserQuota(meta.UserId)
 
 	quota := int(ratio*imageCostRatio*1000) * imageRequest.N
 
@@ -146,7 +92,7 @@ func RelayImageHelper(c *gin.Context, relayMode int) *relaymodel.ErrorWithStatus
 		return openai.ErrorWrapper(err, "new_request_failed", http.StatusInternalServerError)
 	}
 	token := c.Request.Header.Get("Authorization")
-	if channelType == common.ChannelTypeAzure { // Azure authentication
+	if meta.ChannelType == common.ChannelTypeAzure { // Azure authentication
 		token = strings.TrimPrefix(token, "Bearer ")
 		req.Header.Set("api-key", token)
 	} else {
@@ -169,25 +115,25 @@ func RelayImageHelper(c *gin.Context, relayMode int) *relaymodel.ErrorWithStatus
 	if err != nil {
 		return openai.ErrorWrapper(err, "close_request_body_failed", http.StatusInternalServerError)
 	}
-	var textResponse openai.ImageResponse
+	var imageResponse openai.ImageResponse
 
 	defer func(ctx context.Context) {
 		if resp.StatusCode != http.StatusOK {
 			return
 		}
-		err := model.PostConsumeTokenQuota(tokenId, quota)
+		err := model.PostConsumeTokenQuota(meta.TokenId, quota)
 		if err != nil {
 			logger.SysError("error consuming token remain quota: " + err.Error())
 		}
-		err = model.CacheUpdateUserQuota(userId)
+		err = model.CacheUpdateUserQuota(meta.UserId)
 		if err != nil {
 			logger.SysError("error update user quota cache: " + err.Error())
 		}
 		if quota != 0 {
 			tokenName := c.GetString("token_name")
 			logContent := fmt.Sprintf("模型倍率 %.2f，分组倍率 %.2f", modelRatio, groupRatio)
-			model.RecordConsumeLog(ctx, userId, channelId, 0, 0, imageModel, tokenName, quota, logContent)
+			model.RecordConsumeLog(ctx, meta.UserId, meta.ChannelId, 0, 0, imageRequest.Model, tokenName, quota, logContent)
-			model.UpdateUserUsedQuotaAndRequestCount(userId, quota)
+			model.UpdateUserUsedQuotaAndRequestCount(meta.UserId, quota)
 			channelId := c.GetInt("channel_id")
 			model.UpdateChannelUsedQuota(channelId, quota)
 		}
@@ -202,7 +148,7 @@ func RelayImageHelper(c *gin.Context, relayMode int) *relaymodel.ErrorWithStatus
 	if err != nil {
 		return openai.ErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError)
 	}
-	err = json.Unmarshal(responseBody, &textResponse)
+	err = json.Unmarshal(responseBody, &imageResponse)
 	if err != nil {
 		return openai.ErrorWrapper(err, "unmarshal_response_body_failed", http.StatusInternalServerError)
 	}

relay/controller/text.go

@@ -83,11 +83,12 @@ func RelayTextHelper(c *gin.Context) *model.ErrorWithStatusCode {
 		logger.Errorf(ctx, "DoRequest failed: %s", err.Error())
 		return openai.ErrorWrapper(err, "do_request_failed", http.StatusInternalServerError)
 	}
-	meta.IsStream = meta.IsStream || strings.HasPrefix(resp.Header.Get("Content-Type"), "text/event-stream")
+	errorHappened := (resp.StatusCode != http.StatusOK) || (meta.IsStream && resp.Header.Get("Content-Type") == "application/json")
-	if resp.StatusCode != http.StatusOK {
+	if errorHappened {
 		util.ReturnPreConsumedQuota(ctx, preConsumedQuota, meta.TokenId)
 		return util.RelayErrorHandler(resp)
 	}
+	meta.IsStream = meta.IsStream || strings.HasPrefix(resp.Header.Get("Content-Type"), "text/event-stream")
 
 	// do response
 	usage, respErr := adaptor.DoResponse(c, resp, meta)

relay/util/common.go

@@ -27,7 +27,16 @@ func ShouldDisableChannel(err *relaymodel.Error, statusCode int) bool {
 	if statusCode == http.StatusUnauthorized {
 		return true
 	}
-	if err.Type == "insufficient_quota" || err.Code == "invalid_api_key" || err.Code == "account_deactivated" {
+	switch err.Type {
+	case "insufficient_quota":
+		return true
+	// https://docs.anthropic.com/claude/reference/errors
+	case "authentication_error":
+		return true
+	case "permission_error":
+		return true
+	}
+	if err.Code == "invalid_api_key" || err.Code == "account_deactivated" {
 		return true
 	}
 	return false
@@ -101,6 +110,9 @@ func RelayErrorHandler(resp *http.Response) (ErrorWithStatusCode *relaymodel.Err
 	if err != nil {
 		return
 	}
+	if config.DebugEnabled {
+		logger.SysLog(fmt.Sprintf("error happened, status code: %d, response: \n%s", resp.StatusCode, string(responseBody)))
+	}
 	err = resp.Body.Close()
 	if err != nil {
 		return

router/api-router.go

@@ -14,6 +14,7 @@ func SetApiRouter(router *gin.Engine) {
 	apiRouter.Use(middleware.GlobalAPIRateLimit())
 	{
 		apiRouter.GET("/status", controller.GetStatus)
+		apiRouter.GET("/models", middleware.UserAuth(), controller.DashboardListModels)
 		apiRouter.GET("/notice", controller.GetNotice)
 		apiRouter.GET("/about", controller.GetAbout)
 		apiRouter.GET("/home_page_content", controller.GetHomePageContent)
@@ -69,7 +70,7 @@ func SetApiRouter(router *gin.Engine) {
 			channelRoute.GET("/search", controller.SearchChannels)
 			channelRoute.GET("/models", controller.ListModels)
 			channelRoute.GET("/:id", controller.GetChannel)
-			channelRoute.GET("/test", controller.TestAllChannels)
+			channelRoute.GET("/test", controller.TestChannels)
 			channelRoute.GET("/test/:id", controller.TestChannel)
 			channelRoute.GET("/update_balance", controller.UpdateAllChannelsBalance)
 			channelRoute.GET("/update_balance/:id", controller.UpdateChannelBalance)

web/berry/src/constants/ChannelConstants.js

@@ -15,7 +15,7 @@ export const CHANNEL_OPTIONS = {
     key: 3,
     text: 'Azure OpenAI',
     value: 3,
-    color: 'orange'
+    color: 'secondary'
   },
   11: {
     key: 11,
@@ -29,6 +29,12 @@ export const CHANNEL_OPTIONS = {
     value: 24,
     color: 'orange'
   },
+  28: {
+    key: 28,
+    text: 'Mistral AI',
+    value: 28,
+    color: 'orange'
+  },
   15: {
     key: 15,
     text: '百度文心千帆',
@@ -83,6 +89,12 @@ export const CHANNEL_OPTIONS = {
     value: 27,
     color: 'default'
   },
+  29: {
+    key: 29,
+    text: 'Groq',
+    value: 29,
+    color: 'default'
+  },
   8: {
     key: 8,
     text: '自定义渠道',

web/berry/src/views/Channel/type/Config.js

@@ -163,6 +163,9 @@ const typeConfig = {
     },
     modelGroup: "minimax",
   },
+  29: {
+    modelGroup: "groq",
+  },
 };
 
 export { defaultConfig, typeConfig };

web/default/src/components/ChannelsTable.js

@@ -1,7 +1,16 @@
 import React, { useEffect, useState } from 'react';
 import { Button, Form, Input, Label, Message, Pagination, Popup, Table } from 'semantic-ui-react';
 import { Link } from 'react-router-dom';
-import { API, setPromptShown, shouldShowPrompt, showError, showInfo, showSuccess, timestamp2string } from '../helpers';
+import {
+  API,
+  loadChannelModels,
+  setPromptShown,
+  shouldShowPrompt,
+  showError,
+  showInfo,
+  showSuccess,
+  timestamp2string
+} from '../helpers';
 
 import { CHANNEL_OPTIONS, ITEMS_PER_PAGE } from '../constants';
 import { renderGroup, renderNumber } from '../helpers/render';
@@ -95,6 +104,7 @@ const ChannelsTable = () => {
       .catch((reason) => {
         showError(reason);
       });
+    loadChannelModels().then();
   }, []);
 
   const manageChannel = async (id, action, idx, value) => {
@@ -230,11 +240,11 @@ const ChannelsTable = () => {
     }
   };
 
-  const testAllChannels = async () => {
+  const testChannels = async (scope) => {
-    const res = await API.get(`/api/channel/test`);
+    const res = await API.get(`/api/channel/test?scope=${scope}`);
     const { success, message } = res.data;
     if (success) {
-      showInfo('已成功开始测试所有通道，请刷新页面查看结果。');
+      showInfo('已成功开始测试通道，请刷新页面查看结果。');
     } else {
       showError(message);
     }
@@ -519,9 +529,12 @@ const ChannelsTable = () => {
               <Button size='small' as={Link} to='/channel/add' loading={loading}>
                 添加新的渠道
               </Button>
-              <Button size='small' loading={loading} onClick={testAllChannels}>
+              <Button size='small' loading={loading} onClick={()=>{testChannels("all")}}>
                 测试所有渠道
               </Button>
+              <Button size='small' loading={loading} onClick={()=>{testChannels("disabled")}}>
+                测试禁用渠道
+              </Button>
               {/*<Button size='small' onClick={updateAllChannelsBalance}*/}
               {/*        loading={loading || updatingBalance}>更新已启用渠道余额</Button>*/}
               <Popup

web/default/src/components/PasswordResetForm.js

@@ -16,6 +16,17 @@ const PasswordResetForm = () => {
   const [disableButton, setDisableButton] = useState(false);
   const [countdown, setCountdown] = useState(30);
 
+  useEffect(() => {
+    let status = localStorage.getItem('status');
+    if (status) {
+      status = JSON.parse(status);
+      if (status.turnstile_check) {
+        setTurnstileEnabled(true);
+        setTurnstileSiteKey(status.turnstile_site_key);
+      }
+    }
+  }, []);
+
   useEffect(() => {
     let countdownInterval = null;
     if (disableButton && countdown > 0) {

web/default/src/components/SystemSetting.js

@@ -22,6 +22,8 @@ const SystemSetting = () => {
     WeChatServerAddress: '',
     WeChatServerToken: '',
     WeChatAccountQRCodeImageURL: '',
+    MessagePusherAddress: '',
+    MessagePusherToken: '',
     TurnstileCheckEnabled: '',
     TurnstileSiteKey: '',
     TurnstileSecretKey: '',
@@ -183,6 +185,21 @@ const SystemSetting = () => {
     }
   };
 
+  const submitMessagePusher = async () => {
+    if (originInputs['MessagePusherAddress'] !== inputs.MessagePusherAddress) {
+      await updateOption(
+        'MessagePusherAddress',
+        removeTrailingSlash(inputs.MessagePusherAddress)
+      );
+    }
+    if (
+      originInputs['MessagePusherToken'] !== inputs.MessagePusherToken &&
+      inputs.MessagePusherToken !== ''
+    ) {
+      await updateOption('MessagePusherToken', inputs.MessagePusherToken);
+    }
+  };
+
   const submitGitHubOAuth = async () => {
     if (originInputs['GitHubClientId'] !== inputs.GitHubClientId) {
       await updateOption('GitHubClientId', inputs.GitHubClientId);
@@ -496,6 +513,42 @@ const SystemSetting = () => {
             保存 WeChat Server 设置
           </Form.Button>
           <Divider />
+          <Header as='h3'>
+            配置 Message Pusher
+            <Header.Subheader>
+              用以推送报警信息，
+              <a
+                href='https://github.com/songquanpeng/message-pusher'
+                target='_blank'
+              >
+                点击此处
+              </a>
+              了解 Message Pusher
+            </Header.Subheader>
+          </Header>
+          <Form.Group widths={3}>
+            <Form.Input
+              label='Message Pusher 推送地址'
+              name='MessagePusherAddress'
+              placeholder='例如：https://msgpusher.com/push/your_username'
+              onChange={handleInputChange}
+              autoComplete='new-password'
+              value={inputs.MessagePusherAddress}
+            />
+            <Form.Input
+              label='Message Pusher 访问凭证'
+              name='MessagePusherToken'
+              type='password'
+              onChange={handleInputChange}
+              autoComplete='new-password'
+              value={inputs.MessagePusherToken}
+              placeholder='敏感信息不会发送到前端显示'
+            />
+          </Form.Group>
+          <Form.Button onClick={submitMessagePusher}>
+            保存 Message Pusher 设置
+          </Form.Button>
+          <Divider />
           <Header as='h3'>
             配置 Turnstile
             <Header.Subheader>

web/default/src/constants/channel.constants.js

@@ -4,6 +4,7 @@ export const CHANNEL_OPTIONS = [
   { key: 3, text: 'Azure OpenAI', value: 3, color: 'olive' },
   { key: 11, text: 'Google PaLM2', value: 11, color: 'orange' },
   { key: 24, text: 'Google Gemini', value: 24, color: 'orange' },
+  { key: 28, text: 'Mistral AI', value: 28, color: 'orange' },
   { key: 15, text: '百度文心千帆', value: 15, color: 'blue' },
   { key: 17, text: '阿里通义千问', value: 17, color: 'orange' },
   { key: 18, text: '讯飞星火认知', value: 18, color: 'blue' },
@@ -13,6 +14,7 @@ export const CHANNEL_OPTIONS = [
   { key: 23, text: '腾讯混元', value: 23, color: 'teal' },
   { key: 26, text: '百川大模型', value: 26, color: 'orange' },
   { key: 27, text: 'MiniMax', value: 27, color: 'red' },
+  { key: 29, text: 'Groq', value: 29, color: 'orange' },
   { key: 8, text: '自定义渠道', value: 8, color: 'pink' },
   { key: 22, text: '知识库：FastGPT', value: 22, color: 'blue' },
   { key: 21, text: '知识库：AI Proxy', value: 21, color: 'purple' },

web/default/src/helpers/utils.js

@@ -1,11 +1,13 @@
 import { toast } from 'react-toastify';
 import { toastConstants } from '../constants';
 import React from 'react';
+import { API } from './api';
 
 const HTMLToastContent = ({ htmlContent }) => {
   return <div dangerouslySetInnerHTML={{ __html: htmlContent }} />;
 };
 export default HTMLToastContent;
+
 export function isAdmin() {
   let user = localStorage.getItem('user');
   if (!user) return false;
@@ -29,7 +31,7 @@ export function getSystemName() {
 export function getLogo() {
   let logo = localStorage.getItem('logo');
   if (!logo) return '/logo.png';
-  return logo
+  return logo;
 }
 
 export function getFooterHTML() {
@@ -197,3 +199,29 @@ export function shouldShowPrompt(id) {
 export function setPromptShown(id) {
   localStorage.setItem(`prompt-${id}`, 'true');
 }
+
+let channelModels = undefined;
+export async function loadChannelModels() {
+  const res = await API.get('/api/models');
+  const { success, data } = res.data;
+  if (!success) {
+    return;
+  }
+  channelModels = data;
+  localStorage.setItem('channel_models', JSON.stringify(data));
+}
+
+export function getChannelModels(type) {
+  if (channelModels !== undefined && type in channelModels) {
+    return channelModels[type];
+  }
+  let models = localStorage.getItem('channel_models');
+  if (!models) {
+    return [];
+  }
+  channelModels = JSON.parse(models);
+  if (type in channelModels) {
+    return channelModels[type];
+  }
+  return [];
+}

web/default/src/pages/Channel/EditChannel.js

@@ -1,7 +1,7 @@
 import React, { useEffect, useState } from 'react';
 import { Button, Form, Header, Input, Message, Segment } from 'semantic-ui-react';
 import { useNavigate, useParams } from 'react-router-dom';
-import { API, showError, showInfo, showSuccess, verifyJSON } from '../../helpers';
+import { API, copy, getChannelModels, showError, showInfo, showSuccess, verifyJSON } from '../../helpers';
 import { CHANNEL_OPTIONS } from '../../constants';
 
 const MODEL_MAPPING_EXAMPLE = {
@@ -56,60 +56,12 @@ const EditChannel = () => {
   const [customModel, setCustomModel] = useState('');
   const handleInputChange = (e, { name, value }) => {
     setInputs((inputs) => ({ ...inputs, [name]: value }));
-    if (name === 'type' && inputs.models.length === 0) {
+    if (name === 'type') {
-      let localModels = [];
+      let localModels = getChannelModels(value);
-      switch (value) {
+      if (inputs.models.length === 0) {
-        case 14:
+        setInputs((inputs) => ({ ...inputs, models: localModels }));
-          localModels = ['claude-instant-1', 'claude-2', 'claude-2.0', 'claude-2.1'];
-          break;
-        case 11:
-          localModels = ['PaLM-2'];
-          break;
-        case 15:
-          localModels = ['ERNIE-Bot', 'ERNIE-Bot-turbo', 'ERNIE-Bot-4', 'Embedding-V1'];
-          break;
-        case 17:
-          localModels = ['qwen-turbo', 'qwen-plus', 'qwen-max', 'qwen-max-longcontext', 'text-embedding-v1'];
-          let withInternetVersion = [];
-          for (let i = 0; i < localModels.length; i++) {
-            if (localModels[i].startsWith('qwen-')) {
-              withInternetVersion.push(localModels[i] + '-internet');
-            }
-          }
-          localModels = [...localModels, ...withInternetVersion];
-          break;
-        case 16:
-          localModels = ["glm-4", "glm-4v", "glm-3-turbo",'chatglm_turbo', 'chatglm_pro', 'chatglm_std', 'chatglm_lite'];
-          break;
-        case 18:
-          localModels = [
-            'SparkDesk',
-            'SparkDesk-v1.1',
-            'SparkDesk-v2.1',
-            'SparkDesk-v3.1',
-            'SparkDesk-v3.5'
-          ];
-          break;
-        case 19:
-          localModels = ['360GPT_S2_V9', 'embedding-bert-512-v1', 'embedding_s1_v1', 'semantic_similarity_s1_v1'];
-          break;
-        case 23:
-          localModels = ['hunyuan'];
-          break;
-        case 24:
-          localModels = ['gemini-pro', 'gemini-pro-vision'];
-          break;
-        case 25:
-          localModels = ['moonshot-v1-8k', 'moonshot-v1-32k', 'moonshot-v1-128k'];
-          break;
-        case 26:
-          localModels = ['Baichuan2-Turbo', 'Baichuan2-Turbo-192k', 'Baichuan-Text-Embedding'];
-          break;
-        case 27:
-          localModels = ['abab5.5s-chat', 'abab5.5-chat', 'abab6-chat'];
-          break;
       }
-      setInputs((inputs) => ({ ...inputs, models: localModels }));
+      setBasicModels(localModels);
     }
   };
 
@@ -262,6 +214,7 @@ const EditChannel = () => {
               label='类型'
               name='type'
               required
+              search
               options={CHANNEL_OPTIONS}
               value={inputs.type}
               onChange={handleInputChange}
@@ -390,6 +343,8 @@ const EditChannel = () => {
               required
               fluid
               multiple
+              search
+              onLabelClick={(e, { value }) => {copy(value).then()}}
               selection
               onChange={handleInputChange}
               value={inputs.models}