Fix SSE headers initialization for geminiStreamGenerateContent and internalStreamGenerateContent

- Added conditional logic to properly initialize SSE headers only when `alt` is empty.
- Ensured headers like `Content-Type`, `Cache-Control`, and `Access-Control-Allow-Origin` are set for better compatibility.

.github/workflows/docker-image.yml

@@ -0,0 +1,42 @@
+name: docker-image
+
+on:
+  push:
+    tags:
+      - v*
+
+env:
+  APP_NAME: CLIProxyAPI
+  DOCKERHUB_REPO: eceasy/cli-proxy-api
+
+jobs:
+  docker:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+      - name: Set up QEMU
+        uses: docker/setup-qemu-action@v3
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@v3
+      - name: Login to DockerHub
+        uses: docker/login-action@v3
+        with:
+          username: ${{ secrets.DOCKERHUB_USERNAME }}
+          password: ${{ secrets.DOCKERHUB_TOKEN }}
+      - name: Generate App Version
+        run: echo APP_VERSION=`git describe --tags --always` >> $GITHUB_ENV
+      - name: Build and push
+        uses: docker/build-push-action@v6
+        with:
+          context: .
+          platforms: |
+            linux/amd64
+            linux/arm64
+          push: true
+          build-args: |
+            APP_NAME=${{ env.APP_NAME }}
+            APP_VERSION=${{ env.APP_VERSION }}
+          tags: |
+            ${{ env.DOCKERHUB_REPO }}:latest
+            ${{ env.DOCKERHUB_REPO }}:${{ env.APP_VERSION }}

.goreleaser.yml

@@ -14,4 +14,5 @@ archives:
     files:
       - LICENSE
       - README.md
+      - README_CN.md
       - config.yaml

Dockerfile

@@ -0,0 +1,23 @@
+FROM golang:1.24-alpine AS builder
+
+WORKDIR /app
+
+COPY go.mod go.sum ./
+
+RUN go mod download
+
+COPY . .
+
+RUN CGO_ENABLED=0 GOOS=linux go build -o ./CLIProxyAPI ./cmd/server/
+
+FROM alpine:3.22.0
+
+RUN mkdir /CLIProxyAPI
+
+COPY --from=builder ./app/CLIProxyAPI /CLIProxyAPI/CLIProxyAPI
+
+WORKDIR /CLIProxyAPI
+
+EXPOSE 8317
+
+CMD ["./CLIProxyAPI"]

README.md

@@ -1,5 +1,7 @@
 # CLI Proxy API
 
+English | [中文](README_CN.md)
+
 A proxy server that provides an OpenAI/Gemini/Claude compatible API interface for CLI. This allows you to use CLI models with tools and libraries designed for the OpenAI/Gemini/Claude API.
 
 ## Features
@@ -148,14 +150,17 @@ The server uses a YAML configuration file (`config.yaml`) located in the project
 
 ### Configuration Options
 
-| Parameter                     | Type     | Default            | Description                                                                                  |
-|-------------------------------|----------|--------------------|----------------------------------------------------------------------------------------------|
-| `port`                        | integer  | 8317               | The port number on which the server will listen                                              |
-| `auth-dir`                    | string   | "~/.cli-proxy-api" | Directory where authentication tokens are stored. Supports using `~` for home directory      |
-| `proxy-url`                   | string   | ""                 | Proxy url, support socks5/http/https protocol, example: socks5://user:pass@192.168.1.1:1080/ |
-| `debug`                       | boolean  | false              | Enable debug mode for verbose logging                                                        |
-| `api-keys`                    | string[] | []                 | List of API keys that can be used to authenticate requests                                   |
-| `generative-language-api-key` | string[] | []                 | List of Generative Language API keys                                                         |
+| Parameter                             | Type     | Default            | Description                                                                                  |
+|---------------------------------------|----------|--------------------|----------------------------------------------------------------------------------------------|
+| `port`                                | integer  | 8317               | The port number on which the server will listen                                              |
+| `auth-dir`                            | string   | "~/.cli-proxy-api" | Directory where authentication tokens are stored. Supports using `~` for home directory      |
+| `proxy-url`                           | string   | ""                 | Proxy url, support socks5/http/https protocol, example: socks5://user:pass@192.168.1.1:1080/ |
+| `quota-exceeded`                      | object   | {}                 | Configuration for handling quota exceeded                                                    |
+| `quota-exceeded.switch-project`       | boolean  | true               | Whether to automatically switch to another project when a quota is exceeded                  |
+| `quota-exceeded.switch-preview-model` | boolean  | true               | Whether to automatically switch to a preview model when a quota is exceeded                  |
+| `debug`                               | boolean  | false              | Enable debug mode for verbose logging                                                        |
+| `api-keys`                            | string[] | []                 | List of API keys that can be used to authenticate requests                                   |
+| `generative-language-api-key`         | string[] | []                 | List of Generative Language API keys                                                         |
 
 ### Example Configuration File
 
@@ -169,6 +174,14 @@ auth-dir: "~/.cli-proxy-api"
 # Enable debug logging
 debug: false
 
+# Proxy url, support socks5/http/https protocol, example: socks5://user:pass@192.168.1.1:1080/
+proxy-url: ""
+
+# Quota exceeded behavior
+quota-exceeded:
+   switch-project: true # Whether to automatically switch to another project when a quota is exceeded
+   switch-preview-model: true # Whether to automatically switch to a preview model when a quota is exceeded
+
 # API keys for authentication
 api-keys:
   - "your-api-key-1"
@@ -212,6 +225,20 @@ The server will relay the `loadCodeAssist`, `onboardUser`, and `countTokens` req
 > This feature only allows local access because I couldn't find a way to authenticate the requests.   
 > I hardcoded `127.0.0.1` into the load balancing.
 
+## Run with Docker
+
+Run the following command to login: 
+
+```bash
+docker run --rm -p 8085:8085 -v /path/to/your/config.yaml:/CLIProxyAPI/config.yaml -v /path/to/your/auth-dir:/root/.cli-proxy-api eceasy/cli-proxy-api:latest /CLIProxyAPI/CLIProxyAPI --login
+```
+
+Run the following command to start the server:
+
+```bash
+docker run --rm -p 8317:8317 -v /path/to/your/config.yaml:/CLIProxyAPI/config.yaml -v /path/to/your/auth-dir:/root/.cli-proxy-api eceasy/cli-proxy-api:latest
+```
+
 ## Contributing
 
 Contributions are welcome! Please feel free to submit a Pull Request.

README_CN.md

@@ -0,0 +1,254 @@
+# CLI 代理 API
+
+[English](README.md) | 中文
+
+一个为 CLI 提供 OpenAI/Gemini/Claude 兼容 API 接口的代理服务器。这让您可以摆脱终端界面的束缚，将 Gemini 的强大能力以 API 的形式轻松接入到任何您喜爱的客户端或应用中。
+
+## 功能特性
+
+- 为 CLI 模型提供 OpenAI/Gemini/Claude 兼容的 API 端点
+- 支持流式和非流式响应
+- 函数调用/工具支持
+- 多模态输入支持（文本和图像）
+- 多账户支持与负载均衡
+- 简单的 CLI 身份验证流程
+- 支持 Gemini AIStudio API 密钥
+- 支持 Gemini CLI 多账户轮询
+
+## 安装
+
+### 前置要求
+
+- Go 1.24 或更高版本
+- 有权访问 CLI 模型的 Google 账户
+
+### 从源码构建
+
+1. 克隆仓库：
+   ```bash
+   git clone https://github.com/luispater/CLIProxyAPI.git
+   cd CLIProxyAPI
+   ```
+
+2. 构建应用程序：
+   ```bash
+   go build -o cli-proxy-api ./cmd/server
+   ```
+
+## 使用方法
+
+### 身份验证
+
+在使用 API 之前，您需要使用 Google 账户进行身份验证：
+
+```bash
+./cli-proxy-api --login
+```
+
+如果您是旧版 gemini code 用户，可能需要指定项目 ID：
+
+```bash
+./cli-proxy-api --login --project_id <your_project_id>
+```
+
+### 启动服务器
+
+身份验证完成后，启动服务器：
+
+```bash
+./cli-proxy-api
+```
+
+默认情况下，服务器在端口 8317 上运行。
+
+### API 端点
+
+#### 列出模型
+
+```
+GET http://localhost:8317/v1/models
+```
+
+#### 聊天补全
+
+```
+POST http://localhost:8317/v1/chat/completions
+```
+
+请求体示例：
+
+```json
+{
+  "model": "gemini-2.5-pro",
+  "messages": [
+    {
+      "role": "user",
+      "content": "你好，你好吗？"
+    }
+  ],
+  "stream": true
+}
+```
+
+### 与 OpenAI 库一起使用
+
+您可以通过将基础 URL 设置为本地服务器来将此代理与任何 OpenAI 兼容的库一起使用：
+
+#### Python（使用 OpenAI 库）
+
+```python
+from openai import OpenAI
+
+client = OpenAI(
+    api_key="dummy",  # 不使用但必需
+    base_url="http://localhost:8317/v1"
+)
+
+response = client.chat.completions.create(
+    model="gemini-2.5-pro",
+    messages=[
+        {"role": "user", "content": "你好，你好吗？"}
+    ]
+)
+
+print(response.choices[0].message.content)
+```
+
+#### JavaScript/TypeScript
+
+```javascript
+import OpenAI from 'openai';
+
+const openai = new OpenAI({
+  apiKey: 'dummy', // 不使用但必需
+  baseURL: 'http://localhost:8317/v1',
+});
+
+const response = await openai.chat.completions.create({
+  model: 'gemini-2.5-pro',
+  messages: [
+    { role: 'user', content: '你好，你好吗？' }
+  ],
+});
+
+console.log(response.choices[0].message.content);
+```
+
+## 支持的模型
+
+- gemini-2.5-pro
+- gemini-2.5-flash
+- 并且自动切换到之前的预览版本
+
+## 配置
+
+服务器默认使用位于项目根目录的 YAML 配置文件（`config.yaml`）。您可以使用 `--config` 标志指定不同的配置文件路径：
+
+```bash
+./cli-proxy --config /path/to/your/config.yaml
+```
+
+### 配置选项
+
+| 参数                                    | 类型       | 默认值                | 描述                                                                     |
+|---------------------------------------|----------|--------------------|------------------------------------------------------------------------|
+| `port`                                | integer  | 8317               | 服务器监听的端口号                                                              |
+| `auth-dir`                            | string   | "~/.cli-proxy-api" | 存储身份验证令牌的目录。支持使用 `~` 表示主目录                                             |
+| `proxy-url`                           | string   | ""                 | 代理 URL，支持 socks5/http/https 协议，示例：socks5://user:pass@192.168.1.1:1080/ |
+| `quota-exceeded`                      | object   | {}                 | 处理配额超限的配置                                                              |
+| `quota-exceeded.switch-project`       | boolean  | true               | 当配额超限时是否自动切换到另一个项目                                                     |
+| `quota-exceeded.switch-preview-model` | boolean  | true               | 当配额超限时是否自动切换到预览模型                                                      |
+| `debug`                               | boolean  | false              | 启用调试模式以进行详细日志记录                                                        |
+| `api-keys`                            | string[] | []                 | 可用于验证请求的 API 密钥列表                                                      |
+| `generative-language-api-key`         | string[] | []                 | 生成式语言 API 密钥列表                                                         |
+
+### 配置文件示例
+
+```yaml
+# 服务器端口
+port: 8317
+
+# 身份验证目录（支持 ~ 表示主目录）
+auth-dir: "~/.cli-proxy-api"
+
+# 启用调试日志
+debug: false
+
+# 代理 URL，支持 socks5/http/https 协议，示例：socks5://user:pass@192.168.1.1:1080/
+proxy-url: ""
+
+# 配额超限行为
+quota-exceeded:
+   switch-project: true # 当配额超限时是否自动切换到另一个项目
+   switch-preview-model: true # 当配额超限时是否自动切换到预览模型
+
+# 用于本地身份验证的 API 密钥
+api-keys:
+  - "your-api-key-1"
+  - "your-api-key-2"
+
+# AIStduio Gemini API 的 API 密钥
+generative-language-api-key:
+  - "AIzaSy...01"
+  - "AIzaSy...02"
+  - "AIzaSy...03"
+  - "AIzaSy...04"
+```
+
+### 身份验证目录
+
+`auth-dir` 参数指定身份验证令牌的存储位置。当您运行登录命令时，应用程序将在此目录中创建包含 Google 账户身份验证令牌的 JSON 文件。多个账户可用于轮询。
+
+### API 密钥
+
+`api-keys` 参数允许您定义可用于验证对代理服务器请求的 API 密钥列表。在向 API 发出请求时，您可以在 `Authorization` 标头中包含其中一个密钥：
+
+```
+Authorization: Bearer your-api-key-1
+```
+
+### 官方生成式语言 API
+
+`generative-language-api-key` 参数允许您定义可用于验证对官方 AIStudio Gemini API 请求的 API 密钥列表。
+
+## Gemini CLI 多账户负载均衡
+
+启动 CLI 代理 API 服务器，然后将 `CODE_ASSIST_ENDPOINT` 环境变量设置为 CLI 代理 API 服务器的 URL。
+
+```bash
+export CODE_ASSIST_ENDPOINT="http://127.0.0.1:8317"
+```
+
+服务器将中继 `loadCodeAssist`、`onboardUser` 和 `countTokens` 请求。并自动在多个账户之间轮询文本生成请求。
+
+> [!NOTE]  
+> 此功能仅允许本地访问，因为找不到一个可以验证请求的方法。   
+> 所以只能强制只有 `127.0.0.1` 可以访问。
+
+## 使用 Docker 运行
+
+运行以下命令进行登录：
+
+```bash
+docker run --rm -p 8085:8085 -v /path/to/your/config.yaml:/CLIProxyAPI/config.yaml -v /path/to/your/auth-dir:/root/.cli-proxy-api eceasy/cli-proxy-api:latest /CLIProxyAPI/CLIProxyAPI --login
+```
+
+运行以下命令启动服务器：
+
+```bash
+docker run --rm -p 8317:8317 -v /path/to/your/config.yaml:/CLIProxyAPI/config.yaml -v /path/to/your/auth-dir:/root/.cli-proxy-api eceasy/cli-proxy-api:latest
+```
+
+## 贡献
+
+欢迎贡献！请随时提交 Pull Request。
+
+1. Fork 仓库
+2. 创建您的功能分支（`git checkout -b feature/amazing-feature`）
+3. 提交您的更改（`git commit -m 'Add some amazing feature'`）
+4. 推送到分支（`git push origin feature/amazing-feature`）
+5. 打开 Pull Request
+
+## 许可证
+
+此项目根据 MIT 许可证授权 - 有关详细信息，请参阅 [LICENSE](LICENSE) 文件。

internal/api/claude-code-handlers.go

@@ -78,62 +78,16 @@ func (h *APIHandlers) ClaudeMessages(c *gin.Context) {
 	// This loop implements a sophisticated load balancing and failover mechanism
 outLoop:
 	for {
-		// Thread-safe client index rotation to distribute load evenly
-		// This ensures fair usage across all available clients
-		mutex.Lock()
-		startIndex := lastUsedClientIndex
-		currentIndex := (startIndex + 1) % len(h.cliClients)
-		lastUsedClientIndex = currentIndex
-		mutex.Unlock()
-
-		// Build a list of available clients, starting from the next client in rotation
-		// This implements round-robin load balancing while filtering out quota-exceeded clients
-		reorderedClients := make([]*client.Client, 0)
-		for i := 0; i < len(h.cliClients); i++ {
-			cliClient = h.cliClients[(startIndex+1+i)%len(h.cliClients)]
-
-			// Skip clients that have exceeded their quota for the requested model
-			if cliClient.IsModelQuotaExceeded(modelName) {
-				// Log different messages based on authentication method (API key vs account)
-				if cliClient.GetGenerativeLanguageAPIKey() == "" {
-					log.Debugf("Model %s is quota exceeded for account %s, project id: %s", modelName, cliClient.GetEmail(), cliClient.GetProjectID())
-				} else {
-					log.Debugf("Model %s is quota exceeded for generative language API Key: %s", modelName, cliClient.GetGenerativeLanguageAPIKey())
-				}
-
-				cliClient = nil
-				continue
-			}
-			reorderedClients = append(reorderedClients, cliClient)
-		}
-
-		// If all clients have exceeded quota, return a 429 Too Many Requests error
-		if len(reorderedClients) == 0 {
-			c.Status(429)
-			_, _ = fmt.Fprint(c.Writer, fmt.Sprintf(`{"error":{"code":429,"message":"All the models of '%s' are quota exceeded","status":"RESOURCE_EXHAUSTED"}}`, modelName))
+		var errorResponse *client.ErrorMessage
+		cliClient, errorResponse = h.getClient(modelName)
+		if errorResponse != nil {
+			c.Status(errorResponse.StatusCode)
+			_, _ = fmt.Fprint(c.Writer, errorResponse.Error)
 			flusher.Flush()
 			cliCancel()
 			return
 		}
 
-		// Attempt to acquire a lock on an available client using non-blocking TryLock
-		// This prevents blocking when a client is busy with another request
-		locked := false
-		for i := 0; i < len(reorderedClients); i++ {
-			cliClient = reorderedClients[i]
-			if cliClient.RequestMutex.TryLock() {
-				locked = true
-				break
-			}
-		}
-
-		// If no client is immediately available, fall back to blocking on the first client
-		// This ensures the request will eventually be processed
-		if !locked {
-			cliClient = h.cliClients[0]
-			cliClient.RequestMutex.Lock()
-		}
-
 		// Determine the authentication method being used by the selected client
 		// This affects how responses are formatted and logged
 		isGlAPIKey := false

internal/api/cli-handlers.go

@@ -99,6 +99,15 @@ func (h *APIHandlers) CLIHandler(c *gin.Context) {
 }
 
 func (h *APIHandlers) internalStreamGenerateContent(c *gin.Context, rawJson []byte) {
+	alt := h.getAlt(c)
+
+	if alt == "" {
+		c.Header("Content-Type", "text/event-stream")
+		c.Header("Cache-Control", "no-cache")
+		c.Header("Connection", "keep-alive")
+		c.Header("Access-Control-Allow-Origin", "*")
+	}
+
 	// Get the http.Flusher interface to manually flush the response.
 	flusher, ok := c.Writer.(http.Flusher)
 	if !ok {
@@ -141,7 +150,7 @@ outLoop:
 			log.Debugf("Request use account: %s, project id: %s", cliClient.GetEmail(), cliClient.GetProjectID())
 		}
 		// Send the message and receive response chunks and errors via channels.
-		respChan, errChan := cliClient.SendRawMessageStream(cliCtx, rawJson)
+		respChan, errChan := cliClient.SendRawMessageStream(cliCtx, rawJson, "")
 		hasFirstResponse := false
 		for {
 			select {
@@ -220,7 +229,7 @@ func (h *APIHandlers) internalGenerateContent(c *gin.Context, rawJson []byte) {
 			log.Debugf("Request use account: %s, project id: %s", cliClient.GetEmail(), cliClient.GetProjectID())
 		}
 
-		resp, err := cliClient.SendRawMessage(cliCtx, rawJson)
+		resp, err := cliClient.SendRawMessage(cliCtx, rawJson, "")
 		if err != nil {
 			if err.StatusCode == 429 && h.cfg.QuotaExceeded.SwitchProject {
 				continue

internal/api/gemini-handlers.go

@@ -14,11 +14,27 @@ import (
 	"time"
 )
 
-func (h *APIHandlers) GeminiHandler(c *gin.Context) {
-	var person struct {
+func (h *APIHandlers) GeminiModels(c *gin.Context) {
+	c.Status(http.StatusOK)
+	c.Header("Content-Type", "application/json; charset=UTF-8")
+	_, _ = c.Writer.Write([]byte(`{"models":[{"name":"models/gemini-2.5-flash","version":"001","displayName":"Gemini `))
+	_, _ = c.Writer.Write([]byte(`2.5 Flash","description":"Stable version of Gemini 2.5 Flash, our mid-size multimod`))
+	_, _ = c.Writer.Write([]byte(`al model that supports up to 1 million tokens, released in June of 2025.","inputTok`))
+	_, _ = c.Writer.Write([]byte(`enLimit":1048576,"outputTokenLimit":65536,"supportedGenerationMethods":["generateCo`))
+	_, _ = c.Writer.Write([]byte(`ntent","countTokens","createCachedContent","batchGenerateContent"],"temperature":1,`))
+	_, _ = c.Writer.Write([]byte(`"topP":0.95,"topK":64,"maxTemperature":2,"thinking":true},{"name":"models/gemini-2.`))
+	_, _ = c.Writer.Write([]byte(`5-pro","version":"2.5","displayName":"Gemini 2.5 Pro","description":"Stable release`))
+	_, _ = c.Writer.Write([]byte(` (June 17th, 2025) of Gemini 2.5 Pro","inputTokenLimit":1048576,"outputTokenLimit":`))
+	_, _ = c.Writer.Write([]byte(`65536,"supportedGenerationMethods":["generateContent","countTokens","createCachedCo`))
+	_, _ = c.Writer.Write([]byte(`ntent","batchGenerateContent"],"temperature":1,"topP":0.95,"topK":64,"maxTemperatur`))
+	_, _ = c.Writer.Write([]byte(`e":2,"thinking":true}],"nextPageToken":""}`))
+}
+
+func (h *APIHandlers) GeminiGetHandler(c *gin.Context) {
+	var request struct {
 		Action string `uri:"action" binding:"required"`
 	}
-	if err := c.ShouldBindUri(&person); err != nil {
+	if err := c.ShouldBindUri(&request); err != nil {
 		c.JSON(http.StatusBadRequest, ErrorResponse{
 			Error: ErrorDetail{
 				Message: fmt.Sprintf("Invalid request: %v", err),
@@ -27,7 +43,45 @@ func (h *APIHandlers) GeminiHandler(c *gin.Context) {
 		})
 		return
 	}
-	action := strings.Split(person.Action, ":")
+	if request.Action == "gemini-2.5-pro" {
+		c.Status(http.StatusOK)
+		c.Header("Content-Type", "application/json; charset=UTF-8")
+		_, _ = c.Writer.Write([]byte(`{"name":"models/gemini-2.5-pro","version":"2.5","displayName":"Gemini 2.5 Pro",`))
+		_, _ = c.Writer.Write([]byte(`"description":"Stable release (June 17th, 2025) of Gemini 2.5 Pro","inputTokenL`))
+		_, _ = c.Writer.Write([]byte(`imit":1048576,"outputTokenLimit":65536,"supportedGenerationMethods":["generateC`))
+		_, _ = c.Writer.Write([]byte(`ontent","countTokens","createCachedContent","batchGenerateContent"],"temperatur`))
+		_, _ = c.Writer.Write([]byte(`e":1,"topP":0.95,"topK":64,"maxTemperature":2,"thinking":true}`))
+	} else if request.Action == "gemini-2.5-flash" {
+		c.Status(http.StatusOK)
+		c.Header("Content-Type", "application/json; charset=UTF-8")
+		_, _ = c.Writer.Write([]byte(`{"name":"models/gemini-2.5-flash","version":"001","displayName":"Gemini 2.5 Fla`))
+		_, _ = c.Writer.Write([]byte(`sh","description":"Stable version of Gemini 2.5 Flash, our mid-size multimodal `))
+		_, _ = c.Writer.Write([]byte(`model that supports up to 1 million tokens, released in June of 2025.","inputTo`))
+		_, _ = c.Writer.Write([]byte(`kenLimit":1048576,"outputTokenLimit":65536,"supportedGenerationMethods":["gener`))
+		_, _ = c.Writer.Write([]byte(`ateContent","countTokens","createCachedContent","batchGenerateContent"],"temper`))
+		_, _ = c.Writer.Write([]byte(`ature":1,"topP":0.95,"topK":64,"maxTemperature":2,"thinking":true}`))
+	} else {
+		c.Status(http.StatusNotFound)
+		_, _ = c.Writer.Write([]byte(
+			`{"error":{"message":"Not Found","code":404,"status":"NOT_FOUND"}}`,
+		))
+	}
+}
+
+func (h *APIHandlers) GeminiHandler(c *gin.Context) {
+	var request struct {
+		Action string `uri:"action" binding:"required"`
+	}
+	if err := c.ShouldBindUri(&request); err != nil {
+		c.JSON(http.StatusBadRequest, ErrorResponse{
+			Error: ErrorDetail{
+				Message: fmt.Sprintf("Invalid request: %v", err),
+				Type:    "invalid_request_error",
+			},
+		})
+		return
+	}
+	action := strings.Split(request.Action, ":")
 	if len(action) != 2 {
 		c.JSON(http.StatusNotFound, ErrorResponse{
 			Error: ErrorDetail{
@@ -47,10 +101,21 @@ func (h *APIHandlers) GeminiHandler(c *gin.Context) {
 		h.geminiGenerateContent(c, rawJson)
 	} else if method == "streamGenerateContent" {
 		h.geminiStreamGenerateContent(c, rawJson)
+	} else if method == "countTokens" {
+		h.geminiCountTokens(c, rawJson)
 	}
 }
 
 func (h *APIHandlers) geminiStreamGenerateContent(c *gin.Context, rawJson []byte) {
+	alt := h.getAlt(c)
+
+	if alt == "" {
+		c.Header("Content-Type", "text/event-stream")
+		c.Header("Cache-Control", "no-cache")
+		c.Header("Connection", "keep-alive")
+		c.Header("Access-Control-Allow-Origin", "*")
+	}
+
 	// Get the http.Flusher interface to manually flush the response.
 	flusher, ok := c.Writer.(http.Flusher)
 	if !ok {
@@ -118,7 +183,7 @@ outLoop:
 		}
 
 		// Send the message and receive response chunks and errors via channels.
-		respChan, errChan := cliClient.SendRawMessageStream(cliCtx, rawJson)
+		respChan, errChan := cliClient.SendRawMessageStream(cliCtx, rawJson, alt)
 		for {
 			select {
 			// Handle client disconnection.
@@ -135,14 +200,33 @@ outLoop:
 					return
 				} else {
 					if cliClient.GetGenerativeLanguageAPIKey() == "" {
-						responseResult := gjson.GetBytes(chunk, "response")
-						if responseResult.Exists() {
-							chunk = []byte(responseResult.Raw)
+						if alt == "" {
+							responseResult := gjson.GetBytes(chunk, "response")
+							if responseResult.Exists() {
+								chunk = []byte(responseResult.Raw)
+							}
+						} else {
+							chunkTemplate := "[]"
+							responseResult := gjson.ParseBytes(chunk)
+							if responseResult.IsArray() {
+								responseResultItems := responseResult.Array()
+								for i := 0; i < len(responseResultItems); i++ {
+									responseResultItem := responseResultItems[i]
+									if responseResultItem.Get("response").Exists() {
+										chunkTemplate, _ = sjson.SetRaw(chunkTemplate, "-1", responseResultItem.Get("response").Raw)
+									}
+								}
+							}
+							chunk = []byte(chunkTemplate)
 						}
 					}
-					_, _ = c.Writer.Write([]byte("data: "))
-					_, _ = c.Writer.Write(chunk)
-					_, _ = c.Writer.Write([]byte("\n\n"))
+					if alt == "" {
+						_, _ = c.Writer.Write([]byte("data: "))
+						_, _ = c.Writer.Write(chunk)
+						_, _ = c.Writer.Write([]byte("\n\n"))
+					} else {
+						_, _ = c.Writer.Write(chunk)
+					}
 					flusher.Flush()
 				}
 			// Handle errors from the backend.
@@ -165,9 +249,79 @@ outLoop:
 	}
 }
 
+func (h *APIHandlers) geminiCountTokens(c *gin.Context, rawJson []byte) {
+	c.Header("Content-Type", "application/json")
+
+	alt := h.getAlt(c)
+	// orgRawJson := rawJson
+	modelResult := gjson.GetBytes(rawJson, "model")
+	modelName := modelResult.String()
+	cliCtx, cliCancel := context.WithCancel(context.Background())
+	var cliClient *client.Client
+	defer func() {
+		if cliClient != nil {
+			cliClient.RequestMutex.Unlock()
+		}
+	}()
+
+	for {
+		var errorResponse *client.ErrorMessage
+		cliClient, errorResponse = h.getClient(modelName)
+		if errorResponse != nil {
+			c.Status(errorResponse.StatusCode)
+			_, _ = fmt.Fprint(c.Writer, errorResponse.Error)
+			cliCancel()
+			return
+		}
+
+		if glAPIKey := cliClient.GetGenerativeLanguageAPIKey(); glAPIKey != "" {
+			log.Debugf("Request use generative language API Key: %s", glAPIKey)
+		} else {
+			log.Debugf("Request use account: %s, project id: %s", cliClient.GetEmail(), cliClient.GetProjectID())
+
+			template := `{"request":{}}`
+			if gjson.GetBytes(rawJson, "generateContentRequest").Exists() {
+				template, _ = sjson.SetRaw(template, "request", gjson.GetBytes(rawJson, "generateContentRequest").Raw)
+				template, _ = sjson.Delete(template, "generateContentRequest")
+			} else if gjson.GetBytes(rawJson, "contents").Exists() {
+				template, _ = sjson.SetRaw(template, "request.contents", gjson.GetBytes(rawJson, "contents").Raw)
+				template, _ = sjson.Delete(template, "contents")
+			}
+			rawJson = []byte(template)
+		}
+
+		resp, err := cliClient.SendRawTokenCount(cliCtx, rawJson, alt)
+		if err != nil {
+			if err.StatusCode == 429 && h.cfg.QuotaExceeded.SwitchProject {
+				continue
+			} else {
+				c.Status(err.StatusCode)
+				_, _ = c.Writer.Write([]byte(err.Error.Error()))
+				cliCancel()
+				// log.Debugf(err.Error.Error())
+				// log.Debugf(string(rawJson))
+				// log.Debugf(string(orgRawJson))
+			}
+			break
+		} else {
+			if cliClient.GetGenerativeLanguageAPIKey() == "" {
+				responseResult := gjson.GetBytes(resp, "response")
+				if responseResult.Exists() {
+					resp = []byte(responseResult.Raw)
+				}
+			}
+			_, _ = c.Writer.Write(resp)
+			cliCancel()
+			break
+		}
+	}
+}
+
 func (h *APIHandlers) geminiGenerateContent(c *gin.Context, rawJson []byte) {
 	c.Header("Content-Type", "application/json")
 
+	alt := h.getAlt(c)
+
 	modelResult := gjson.GetBytes(rawJson, "model")
 	modelName := modelResult.String()
 	cliCtx, cliCancel := context.WithCancel(context.Background())
@@ -217,7 +371,7 @@ func (h *APIHandlers) geminiGenerateContent(c *gin.Context, rawJson []byte) {
 		} else {
 			log.Debugf("Request use account: %s, project id: %s", cliClient.GetEmail(), cliClient.GetProjectID())
 		}
-		resp, err := cliClient.SendRawMessage(cliCtx, rawJson)
+		resp, err := cliClient.SendRawMessage(cliCtx, rawJson, alt)
 		if err != nil {
 			if err.StatusCode == 429 && h.cfg.QuotaExceeded.SwitchProject {
 				continue
@@ -240,3 +394,16 @@ func (h *APIHandlers) geminiGenerateContent(c *gin.Context, rawJson []byte) {
 		}
 	}
 }
+
+func (h *APIHandlers) getAlt(c *gin.Context) string {
+	var alt string
+	var hasAlt bool
+	alt, hasAlt = c.GetQuery("alt")
+	if !hasAlt {
+		alt, _ = c.GetQuery("$alt")
+	}
+	if alt == "sse" {
+		return ""
+	}
+	return alt
+}

internal/api/handlers.go

@@ -86,6 +86,10 @@ func (h *APIHandlers) Models(c *gin.Context) {
 }
 
 func (h *APIHandlers) getClient(modelName string) (*client.Client, *client.ErrorMessage) {
+	if len(h.cliClients) == 0 {
+		return nil, &client.ErrorMessage{StatusCode: 500, Error: fmt.Errorf("no clients available")}
+	}
+
 	var cliClient *client.Client
 
 	// Lock the mutex to update the last used client index

internal/api/server.go

@@ -75,8 +75,9 @@ func (s *Server) setupRoutes() {
 	v1beta := s.engine.Group("/v1beta")
 	v1beta.Use(AuthMiddleware(s.cfg))
 	{
-		v1beta.GET("/models", s.handlers.Models)
+		v1beta.GET("/models", s.handlers.GeminiModels)
 		v1beta.POST("/models/:action", s.handlers.GeminiHandler)
+		v1beta.GET("/models/:action", s.handlers.GeminiGetHandler)
 	}
 
 	// Root endpoint
@@ -151,7 +152,11 @@ func AuthMiddleware(cfg *config.Config) gin.HandlerFunc {
 		authHeader := c.GetHeader("Authorization")
 		authHeaderGoogle := c.GetHeader("X-Goog-Api-Key")
 		authHeaderAnthropic := c.GetHeader("X-Api-Key")
-		if authHeader == "" && authHeaderGoogle == "" && authHeaderAnthropic == "" {
+
+		// Get the API key from the query parameter
+		apiKeyQuery, _ := c.GetQuery("key")
+
+		if authHeader == "" && authHeaderGoogle == "" && authHeaderAnthropic == "" && apiKeyQuery == "" {
 			c.AbortWithStatusJSON(http.StatusUnauthorized, gin.H{
 				"error": "Missing API key",
 			})
@@ -170,7 +175,7 @@ func AuthMiddleware(cfg *config.Config) gin.HandlerFunc {
 		// Find the API key in the in-memory list
 		var foundKey string
 		for i := range cfg.ApiKeys {
-			if cfg.ApiKeys[i] == apiKey || cfg.ApiKeys[i] == authHeaderGoogle || cfg.ApiKeys[i] == authHeaderAnthropic {
+			if cfg.ApiKeys[i] == apiKey || cfg.ApiKeys[i] == authHeaderGoogle || cfg.ApiKeys[i] == authHeaderAnthropic || cfg.ApiKeys[i] == apiKeyQuery {
 				foundKey = cfg.ApiKeys[i]
 				break
 			}

internal/auth/auth.go

@@ -168,11 +168,12 @@ func getTokenFromWeb(ctx context.Context, config *oauth2.Config) (*oauth2.Token,
 	codeChan := make(chan string)
 	errChan := make(chan error)
 
-	// Create a new HTTP server.
-	server := &http.Server{Addr: "localhost:8085"}
+	// Create a new HTTP server with its own multiplexer.
+	mux := http.NewServeMux()
+	server := &http.Server{Addr: ":8085", Handler: mux}
 	config.RedirectURL = "http://localhost:8085/oauth2callback"
 
-	http.HandleFunc("/oauth2callback", func(w http.ResponseWriter, r *http.Request) {
+	mux.HandleFunc("/oauth2callback", func(w http.ResponseWriter, r *http.Request) {
 		if err := r.URL.Query().Get("error"); err != "" {
 			_, _ = fmt.Fprintf(w, "Authentication failed: %s", err)
 			errChan <- fmt.Errorf("authentication failed via callback: %s", err)

internal/client/client.go

@@ -28,7 +28,7 @@ const (
 	apiVersion         = "v1internal"
 	pluginVersion      = "0.1.9"
 
-	glEndPoint   = "https://generativelanguage.googleapis.com/"
+	glEndPoint   = "https://generativelanguage.googleapis.com"
 	glApiVersion = "v1beta"
 )
 
@@ -241,7 +241,7 @@ func (c *Client) makeAPIRequest(ctx context.Context, endpoint, method string, bo
 }
 
 // APIRequest handles making requests to the CLI API endpoints.
-func (c *Client) APIRequest(ctx context.Context, endpoint string, body interface{}, stream bool) (io.ReadCloser, *ErrorMessage) {
+func (c *Client) APIRequest(ctx context.Context, endpoint string, body interface{}, alt string, stream bool) (io.ReadCloser, *ErrorMessage) {
 	var jsonBody []byte
 	var err error
 	if byteBody, ok := body.([]byte); ok {
@@ -257,25 +257,39 @@ func (c *Client) APIRequest(ctx context.Context, endpoint string, body interface
 	if c.glAPIKey == "" {
 		// Add alt=sse for streaming
 		url = fmt.Sprintf("%s/%s:%s", codeAssistEndpoint, apiVersion, endpoint)
-		if stream {
+		if alt == "" && stream {
 			url = url + "?alt=sse"
+		} else {
+			if alt != "" {
+				url = url + fmt.Sprintf("?$alt=%s", alt)
+			}
 		}
 	} else {
-		modelResult := gjson.GetBytes(jsonBody, "model")
-		url = fmt.Sprintf("%s/%s/models/%s:%s", glEndPoint, glApiVersion, modelResult.String(), endpoint)
-		if stream {
-			url = url + "?alt=sse"
-		}
-		jsonBody = []byte(gjson.GetBytes(jsonBody, "request").Raw)
-		systemInstructionResult := gjson.GetBytes(jsonBody, "systemInstruction")
-		if systemInstructionResult.Exists() {
-			jsonBody, _ = sjson.SetRawBytes(jsonBody, "system_instruction", []byte(systemInstructionResult.Raw))
-			jsonBody, _ = sjson.DeleteBytes(jsonBody, "systemInstruction")
-			jsonBody, _ = sjson.DeleteBytes(jsonBody, "session_id")
+		if endpoint == "countTokens" {
+			modelResult := gjson.GetBytes(jsonBody, "model")
+			url = fmt.Sprintf("%s/%s/models/%s:%s", glEndPoint, glApiVersion, modelResult.String(), endpoint)
+		} else {
+			modelResult := gjson.GetBytes(jsonBody, "model")
+			url = fmt.Sprintf("%s/%s/models/%s:%s", glEndPoint, glApiVersion, modelResult.String(), endpoint)
+			if alt == "" && stream {
+				url = url + "?alt=sse"
+			} else {
+				if alt != "" {
+					url = url + fmt.Sprintf("?$alt=%s", alt)
+				}
+			}
+			jsonBody = []byte(gjson.GetBytes(jsonBody, "request").Raw)
+			systemInstructionResult := gjson.GetBytes(jsonBody, "systemInstruction")
+			if systemInstructionResult.Exists() {
+				jsonBody, _ = sjson.SetRawBytes(jsonBody, "system_instruction", []byte(systemInstructionResult.Raw))
+				jsonBody, _ = sjson.DeleteBytes(jsonBody, "systemInstruction")
+				jsonBody, _ = sjson.DeleteBytes(jsonBody, "session_id")
+			}
 		}
 	}
 
 	// log.Debug(string(jsonBody))
+	// log.Debug(url)
 	reqBody := bytes.NewBuffer(jsonBody)
 
 	req, err := http.NewRequestWithContext(ctx, "POST", url, reqBody)
@@ -392,7 +406,7 @@ func (c *Client) SendMessage(ctx context.Context, rawJson []byte, model string,
 			}
 		}
 
-		respBody, err := c.APIRequest(ctx, "generateContent", byteRequestBody, false)
+		respBody, err := c.APIRequest(ctx, "generateContent", byteRequestBody, "", false)
 		if err != nil {
 			if err.StatusCode == 429 {
 				now := time.Now()
@@ -544,7 +558,7 @@ func (c *Client) SendMessageStream(ctx context.Context, rawJson []byte, model st
 
 			// Attempt to establish a streaming connection with the API
 			var err *ErrorMessage
-			stream, err = c.APIRequest(ctx, "streamGenerateContent", byteRequestBody, true)
+			stream, err = c.APIRequest(ctx, "streamGenerateContent", byteRequestBody, "", true)
 			if err != nil {
 				// Handle quota exceeded errors by marking the model and potentially retrying
 				if err.StatusCode == 429 {
@@ -593,8 +607,49 @@ func (c *Client) SendMessageStream(ctx context.Context, rawJson []byte, model st
 	return dataChan, errChan
 }
 
+// SendRawTokenCount handles a token count.
+func (c *Client) SendRawTokenCount(ctx context.Context, rawJson []byte, alt string) ([]byte, *ErrorMessage) {
+	modelResult := gjson.GetBytes(rawJson, "model")
+	model := modelResult.String()
+	modelName := model
+	for {
+		if c.isModelQuotaExceeded(modelName) {
+			if c.cfg.QuotaExceeded.SwitchPreviewModel && c.glAPIKey == "" {
+				modelName = c.getPreviewModel(model)
+				if modelName != "" {
+					log.Debugf("Model %s is quota exceeded. Switch to preview model %s", model, modelName)
+					rawJson, _ = sjson.SetBytes(rawJson, "model", modelName)
+					continue
+				}
+			}
+			return nil, &ErrorMessage{
+				StatusCode: 429,
+				Error:      fmt.Errorf(`{"error":{"code":429,"message":"All the models of '%s' are quota exceeded","status":"RESOURCE_EXHAUSTED"}}`, model),
+			}
+		}
+
+		respBody, err := c.APIRequest(ctx, "countTokens", rawJson, alt, false)
+		if err != nil {
+			if err.StatusCode == 429 {
+				now := time.Now()
+				c.modelQuotaExceeded[modelName] = &now
+				if c.cfg.QuotaExceeded.SwitchPreviewModel && c.glAPIKey == "" {
+					continue
+				}
+			}
+			return nil, err
+		}
+		delete(c.modelQuotaExceeded, modelName)
+		bodyBytes, errReadAll := io.ReadAll(respBody)
+		if errReadAll != nil {
+			return nil, &ErrorMessage{StatusCode: 500, Error: errReadAll}
+		}
+		return bodyBytes, nil
+	}
+}
+
 // SendRawMessage handles a single conversational turn, including tool calls.
-func (c *Client) SendRawMessage(ctx context.Context, rawJson []byte) ([]byte, *ErrorMessage) {
+func (c *Client) SendRawMessage(ctx context.Context, rawJson []byte, alt string) ([]byte, *ErrorMessage) {
 	if c.glAPIKey == "" {
 		rawJson, _ = sjson.SetBytes(rawJson, "project", c.GetProjectID())
 	}
@@ -618,7 +673,7 @@ func (c *Client) SendRawMessage(ctx context.Context, rawJson []byte) ([]byte, *E
 			}
 		}
 
-		respBody, err := c.APIRequest(ctx, "generateContent", rawJson, false)
+		respBody, err := c.APIRequest(ctx, "generateContent", rawJson, alt, false)
 		if err != nil {
 			if err.StatusCode == 429 {
 				now := time.Now()
@@ -639,7 +694,7 @@ func (c *Client) SendRawMessage(ctx context.Context, rawJson []byte) ([]byte, *E
 }
 
 // SendRawMessageStream handles a single conversational turn, including tool calls.
-func (c *Client) SendRawMessageStream(ctx context.Context, rawJson []byte) (<-chan []byte, <-chan *ErrorMessage) {
+func (c *Client) SendRawMessageStream(ctx context.Context, rawJson []byte, alt string) (<-chan []byte, <-chan *ErrorMessage) {
 	dataTag := []byte("data: ")
 	errChan := make(chan *ErrorMessage)
 	dataChan := make(chan []byte)
@@ -672,7 +727,7 @@ func (c *Client) SendRawMessageStream(ctx context.Context, rawJson []byte) (<-ch
 				return
 			}
 			var err *ErrorMessage
-			stream, err = c.APIRequest(ctx, "streamGenerateContent", rawJson, true)
+			stream, err = c.APIRequest(ctx, "streamGenerateContent", rawJson, alt, true)
 			if err != nil {
 				if err.StatusCode == 429 {
 					now := time.Now()
@@ -688,21 +743,32 @@ func (c *Client) SendRawMessageStream(ctx context.Context, rawJson []byte) (<-ch
 			break
 		}
 
-		scanner := bufio.NewScanner(stream)
-		for scanner.Scan() {
-			line := scanner.Bytes()
-			if bytes.HasPrefix(line, dataTag) {
-				dataChan <- line[6:]
+		if alt == "" {
+			scanner := bufio.NewScanner(stream)
+			for scanner.Scan() {
+				line := scanner.Bytes()
+				if bytes.HasPrefix(line, dataTag) {
+					dataChan <- line[6:]
+				}
 			}
-		}
 
-		if errScanner := scanner.Err(); errScanner != nil {
-			errChan <- &ErrorMessage{500, errScanner}
-			_ = stream.Close()
-			return
-		}
+			if errScanner := scanner.Err(); errScanner != nil {
+				errChan <- &ErrorMessage{500, errScanner}
+				_ = stream.Close()
+				return
+			}
 
+		} else {
+			data, err := io.ReadAll(stream)
+			if err != nil {
+				errChan <- &ErrorMessage{500, err}
+				_ = stream.Close()
+				return
+			}
+			dataChan <- data
+		}
 		_ = stream.Close()
+
 	}()
 
 	return dataChan, errChan
@@ -754,7 +820,7 @@ func (c *Client) CheckCloudAPIIsEnabled() (bool, error) {
 	// A simple request to test the API endpoint.
 	requestBody := fmt.Sprintf(`{"project":"%s","request":{"contents":[{"role":"user","parts":[{"text":"Be concise. What is the capital of France?"}]}],"generationConfig":{"thinkingConfig":{"include_thoughts":false,"thinkingBudget":0}}},"model":"gemini-2.5-flash"}`, c.tokenStorage.ProjectID)
 
-	stream, err := c.APIRequest(ctx, "streamGenerateContent", []byte(requestBody), true)
+	stream, err := c.APIRequest(ctx, "streamGenerateContent", []byte(requestBody), "", true)
 	if err != nil {
 		// If a 403 Forbidden error occurs, it likely means the API is not enabled.
 		if err.StatusCode == 403 {
@@ -771,6 +837,7 @@ func (c *Client) CheckCloudAPIIsEnabled() (bool, error) {
 					)
 				}
 			}
+			log.Warnf("\n\nPlease copy this message and create an issue.\n\n%s\n\n", errJson)
 			return false, nil
 		}
 		return false, err.Error

internal/cmd/login.go

@@ -73,6 +73,7 @@ func DoLogin(cfg *config.Config, projectID string) {
 		// If the check fails (returns false), the CheckCloudAPIIsEnabled function
 		// will have already printed instructions, so we can just exit.
 		if !isChecked {
+			log.Fatal("Failed to check if Cloud AI API is enabled. If you encounter an error message, please create an issue.")
 			return
 		}
 	}