> ## Documentation Index > Fetch the complete documentation index at: https://docs.geekai.co/llms.txt > Use this file to discover all available pages before exploring further. # Chat Completions Note: For setting the chat model name, refer to [System Supported Chat Model List](https://geekai.dev/models). The request/response parameter structure is fully compatible with [OpenAI](https://platform.openai.com/docs/api-reference/chat/create). When switching models, you only need to modify the corresponding model name. If the model request/response parameters are inconsistent with OpenAI, GeekAI will automatically convert and align them. Except for Baidu ERNIE Bot and iFlytek Spark, all other platforms support function calling (the specific supported models are subject to platform restrictions). When calling the GPTs model API, you need to replace the \* in gpt-4-gizmo-\* with the corresponding GPTs' gizmo\_id, which can be obtained by extracting it from the GPTs URL. For example, for [https://chatgpt.com/g/g-bo0FiWLY7-researchgpt](https://chatgpt.com/g/g-bo0FiWLY7-researchgpt), the corresponding gizmo\_id is g-bo0FiWLY7. The response data structure is fully compatible with OpenAI and builds upon it to add new features for adapting to other models. It provides support for cited links (`citations`), search billing units (`billed_units`), video in message content, image/video input tokens, and inference mode settings (`thinking`). The response structure changes depending on whether it is a streaming output or not. You can refer to the request examples below to make a judgment: ### cURL Request Example ```bash theme={null} curl --location --request POST 'https://geekai.dev/api/v1/chat/completions' \ --header 'Authorization: Bearer {YOUR_GEEKAI_API_KEY}' \ --header 'Content-Type: application/json' \ --data-raw '{ "model": "gpt-4o-mini", "messages": [ { "role": "user", "content": "hi" } ], "stream": false }' ``` ## OpenAPI ````yaml openapi.yaml POST /chat/completions openapi: 3.1.0 info: title: 极客智坊 API文档 description: 极客智坊 API 消息接口规范文档 version: 1.0.0 servers: - url: https://geekai.co/api/v1 description: 国内代理入口 - url: https://geekai.dev/api/v1 description: 海外代理入口 security: - bearerAuth: [] - apiKeyAuth: [] paths: /chat/completions: post: tags: - Chat summary: 对话完成接口 description: 用于创建对话完成的接口，支持多种对话模型，可配置各种参数来控制响应的生成。 operationId: createChatCompletion requestBody: required: true content: application/json: schema: type: object required: - model - messages properties: model: type: string description: 对话模型 default: gpt-5-mini messages: type: array description: 消息列表 items: type: object required: - content - role properties: content: oneOf: - type: string description: 文本格式消息内容 - type: array description: 多模态消息内容 items: oneOf: - $ref: '#/components/schemas/TextContentPart' - $ref: '#/components/schemas/ImageContentPart' - $ref: '#/components/schemas/AudioContentPart' - $ref: '#/components/schemas/VideoContentPart' role: type: string description: 消息角色 enum: - system - user - assistant example: - content: 你是一个助手 role: system - content: 你好 role: user enable_thinking: type: boolean description: 开启推理模式，仅支持切换思考模式的模型支持 default: false thinking: type: object description: 推理模式细节配置 properties: include_thoughts: type: boolean description: 是否显示思考链内容，默认false default: false budget_tokens: type: integer description: 推理模式token预算 default: 16000 reasoning_effort: type: string description: 推理模式推理努力程度，适用于 OpenAI，GPT 5.1 起默认 none enum: - none - minimal - low - medium - high default: medium thinking_level: type: string description: 推理模式思考层级，适用于 Gemini 3 Pro，默认 high enum: - low - high default: high stream: type: boolean description: 是否返回流式响应，默认false default: false background: type: boolean description: 是否开启后台任务模式，默认false，注意 stream 和 background 不能同时为 true default: false enable_search: type: boolean description: 是否启用联网搜索，默认false default: false search_config: type: object description: 联网搜索配置明细，仅智谱清言和不支持内置搜索工具的模型适用 properties: engine: type: string description: 搜索引擎 enum: - glm/search-std - glm/search-pro - glm/search-pro-sogou - glm/search-pro-quark default: glm/search-std prompt: type: string description: 定制搜索结果处理的提示语 intent: type: boolean description: 启用搜索意图 default: false count: type: integer description: 搜索结果数量, 取值范围1-50 default: 10 domain_filter: type: string description: 搜索结果域名过滤白名单 recency_filter: type: string description: 搜索结果时间过滤 enum: - noLimit - oneDay - oneWeek - oneMonth - oneYear default: noLimit content_size: type: string description: 搜索结果摘要长度 enum: - medium - high default: medium return_result: type: boolean description: 是否返回搜索结果 default: false result_sequence: type: string description: 搜索结果在对话响应中的位置，主要用于流式响应，位于真正回答内容之前还是之后 enum: - before - after default: after require_search: type: boolean description: 是否强制要求必须有搜索结果才进行回答 default: false deepable: type: boolean description: 是否开启深度搜索（目前未生效） default: false enable_url_context: type: boolean description: 是否启用 URL Context，默认false，仅部分 Gemini 模型生效 default: false temperature: type: number description: 温度参数，默认为代理模型默认值 max_completion_tokens: type: integer description: 最大输出token数,默认设置为当前模型最大支持输出 example: 4096 json_mode: type: boolean description: 是否设置响应格式为JSON对象，默认false default: false response_format: description: 指定响应输出格式，默认不指定为文本输出，如果设置该配置项会覆盖json_mode设置 oneOf: - $ref: '#/components/schemas/ResponseFormatText' - $ref: '#/components/schemas/ResponseFormatJsonObject' - $ref: '#/components/schemas/ResponseFormatJsonSchema' modalities: type: array description: 输出的多模态类型列表 items: type: string enum: - text - image - audio - video audio: type: object description: 音频输出配置 properties: voice: type: string description: 音色 default: alloy format: type: string description: 音频格式 default: wav image: type: object description: 图像生成配置，仅 Gemini 3 Pro Image 适用 properties: aspect_ratio: type: string description: 图片宽高比 enum: - '1:1' - '4:3' - '3:4' - '16:9' - '9:16' image_size: type: string description: 图像分辨率 enum: - 1K - 2K - 4K default: 1K tools: type: array description: 可调用的工具函数列表 tool_choice: description: 模型在生成响应时应如何选择使用哪个工具（或多个工具） oneOf: - $ref: '#/components/schemas/ToolChoiceOptions' - $ref: '#/components/schemas/ToolChoiceTypes' - $ref: '#/components/schemas/ToolChoiceFunction' tool_config: type: object description: 工具函数调用配置(兼容gemini) properties: functionCallingConfig: type: object description: 函数调用配置 properties: mode: type: string description: 函数调用模式 enum: - AUTO - ANY - NONE - VALIDATED default: AUTO allowedFunctionNames: type: array description: 允许调用的函数名称列表 items: type: string parallel_tool_calls: type: boolean description: 是否并发调用工具函数 default: true image_generation: type: boolean description: 是否强制出图，仅支持对话画图的模型支持 default: false stop: type: array description: 停止生成的触发词列表 items: type: string logprobs: type: boolean description: 是否返回token概率 default: false top_logprobs: type: integer description: 每个位置返回的最可能token数 example: 2 frequency_penalty: type: number description: 频率惩罚系数 default: 0 presence_penalty: type: number description: 存在惩罚系数 default: 0 top_p: type: number description: 核采样阈值 default: 1 top_k: type: integer description: 最高概率采样数 default: 1 seed: type: integer description: 随机数种子 'n': type: integer description: 生成结果数量 default: 1 metadata: type: object description: 自定义元数据 sess_id: type: string description: 第三方应用自行实现的会话ID format: uuid retries: type: integer description: 自动重试次数，默认0 default: 0 responses: '200': description: 成功响应 content: application/json: schema: type: object properties: id: type: string description: 请求ID status: type: string description: 请求状态，只有后台模式才会返回 enum: - pending - running - succeed created: type: integer description: 请求创建Unix时间戳 model: type: string description: 对话使用的模型 object: type: string description: 响应对象类型 choices: type: array description: 生成的对话列表 items: type: object properties: index: type: integer description: 对话索引 message: type: object description: 生成的消息 properties: role: type: string enum: - assistant description: 角色 content: type: string description: 响应内容 reasoning_content: type: string description: 推理内容 image: type: string description: 图片URL/Base64数据 audio: type: object description: 音频数据 properties: id: type: string description: 音频唯一识别码 data: type: string description: Base64编码的音频数据 expires_at: type: integer description: 过期Unix时间戳 transcript: type: string description: 音频转录文本 tool_calls: type: array description: 工具函数调用列表 items: type: object properties: id: type: string description: 工具ID type: type: string enum: - function description: 工具类型 function: type: object description: 工具函数 properties: name: type: string description: 函数名称 arguments: type: string description: 函数参数 finish_reason: type: string enum: - stop - length - tool_calls - content_filter - stop_sequence - error description: 结束原因 logprobs: type: object description: 日志概率 properties: content: type: array description: 消息内容标记列表 items: type: object properties: token: type: string description: 令牌 logprob: type: number description: 对数概率 bytes: type: array description: 表示 UTF-8 字节编码的整数列表，代表该标记 items: type: integer top_logprobs: type: array description: 最有可能的标记及其在此位置的对数概率列表 items: type: object properties: token: type: string description: 令牌 logprob: type: number description: 对数概率 bytes: type: array description: 表示 UTF-8 字节编码的整数列表，代表该标记 items: type: integer usage: type: object description: tokens消耗统计 properties: prompt_tokens: type: integer description: 输入文本token数 completion_tokens: type: integer description: 生成文本token数 total_tokens: type: integer description: 总token数 billed_units: type: integer description: 计费次数（搜索） prompt_tokens_details: type: object description: 输入文本消耗明细 properties: text_tokens: type: integer description: 文本token数 audio_tokens: type: integer description: 音频token数 cached_tokens: type: integer description: 缓存token数 image_tokens: type: integer description: 图片token数 video_tokens: type: integer description: 视频token数 citation_tokens: type: integer description: 引用token数 completion_tokens_details: type: object description: 生成文本消耗明细 properties: text_tokens: type: integer description: 文本token数 audio_tokens: type: integer description: 音频token数 reasoning_tokens: type: integer description: 推理token数 accepted_prediction_tokens: type: integer description: 接受的预测token数 rejected_prediction_tokens: type: integer description: 拒绝的预测token数 citations: type: array description: 引用的文献/链接列表 items: type: string system_fingerprint: description: 系统指纹 type: string '400': $ref: '#/components/responses/ValidationError' '401': $ref: '#/components/responses/Unauthorized' '500': $ref: '#/components/responses/StandardError' components: schemas: TextContentPart: type: object required: - type - text properties: type: type: string enum: - text description: 内容类型 text: type: string description: 文本内容 ImageContentPart: type: object required: - type - image_url properties: type: type: string enum: - image_url description: 内容类型 image_url: type: object required: - url properties: url: type: string description: 图片URL或Base64编码的图片数据 description: 图片URL对象 AudioContentPart: type: object required: - type - input_audio properties: type: type: string enum: - input_audio description: 内容类型 input_audio: type: object required: - data - format properties: data: type: string description: Base64编码的音频数据 format: type: string description: 内容类型，目前仅支持 wav/mp3 格式 description: 音频数据对象 VideoContentPart: type: object required: - type - video_url properties: type: type: string enum: - video_url description: 内容类型 video_url: type: object required: - url properties: url: type: string description: 视频URL description: 视频URL对象 ResponseFormatText: type: object title: Text description: | 默认响应格式。用于生成文本响应。 properties: type: type: string description: 正在定义的响应格式的类型。始终为 `text`。 enum: - text x-stainless-const: true required: - type ResponseFormatJsonObject: type: object title: JSON object description: | JSON 对象响应格式。生成 JSON 响应的旧方法。 建议对支持 `json_schema` 的模型使用它。请注意，模型不会在没有系统消息或用户消息指示的情况下生成 JSON。 properties: type: type: string description: 正在定义的响应格式的类型。始终为 `json_object`。 enum: - json_object x-stainless-const: true required: - type ResponseFormatJsonSchema: type: object title: JSON schema description: | JSON Schema 响应格式。用于生成结构化 JSON 响应。 了解有关[结构化输出](/docs/guides/structured-outputs)的更多信息。 properties: type: type: string description: 正在定义的响应格式的类型。始终为 `json_schema`。 enum: - json_schema x-stainless-const: true json_schema: type: object title: JSON schema description: | 结构化输出配置选项，包括 JSON Schema。 properties: description: type: string description: | 响应格式的描述，用于模型确定如何以该格式进行响应。 name: type: string description: | 响应格式的名称。必须是 a-z、A-Z、0-9，或包含下划线和破折号，且最大长度为 64。 schema: $ref: '#/components/schemas/ResponseFormatJsonSchemaSchema' strict: type: boolean nullable: true default: false description: | 是否在生成输出时启用严格的模式遵循。 如果设置为 true，模型将始终遵循 `schema` 字段中定义的确切模式。 当 `strict` 为 `true` 时，仅支持 JSON Schema 的子集。 要了解更多信息，请阅读 [结构化输出指南](/docs/guides/structured-outputs)。 required: - name required: - type - json_schema ToolChoiceOptions: type: string title: 工具选择模式 description: | 控制模型调用哪个（如果有）工具。 `none` 表示模型将不调用任何工具，而是生成一条消息。 `auto` 表示模型可以选择生成消息或调用一个或多个工具。 `required` 表示模型必须调用一个或多个工具。 enum: - none - auto - required ToolChoiceTypes: type: object title: 托管工具 description: | 表示模型应使用内置工具生成响应。 [了解有关内置工具的更多信息](/docs/guides/tools)。 properties: type: type: string description: | 模型应使用的托管工具的类型。详细了解 [内置工具](/docs/guides/tools)。 允许的值为： - `file_search` - `web_search_preview` - `computer_use_preview` - `code_interpreter` - `mcp` - `image_generation` enum: - file_search - web_search_preview - computer_use_preview - web_search_preview_2025_03_11 - image_generation - code_interpreter - mcp required: - type ToolChoiceFunction: type: object title: 函数工具 description: | 使用此选项强制模型调用特定函数。 properties: type: type: string enum: - function description: 对于函数调用，类型始终为 `function`。 x-stainless-const: true name: type: string description: 要调用的函数的名称。 required: - type - name ResponseFormatJsonSchemaSchema: type: object title: JSON schema description: | 响应格式的模式，描述为 JSON Schema 对象。 了解有关如何构建 JSON Schema 的更多信息 [here](https://json-schema.org/)。 additionalProperties: true Error: type: object required: - code - message properties: code: type: string description: 错误代码 example: invalid_request message: type: string description: 错误信息描述 example: 请求参数不合法 responses: ValidationError: description: 参数验证错误 content: application/json: schema: type: object properties: code: type: string example: validation_error message: type: string example: 参数验证失败 details: type: object additionalProperties: type: string example: field: 错误描述 Unauthorized: description: 未授权 content: application/json: schema: type: object properties: code: type: string example: unauthorized message: type: string example: Invalid API key or token StandardError: description: 标准错误响应 content: application/json: schema: $ref: '#/components/schemas/Error' securitySchemes: bearerAuth: type: http scheme: bearer bearerFormat: JWT description: API认证token apiKeyAuth: type: apiKey in: header name: GeekAI-API-Key description: API密钥认证 ````