# 兔子API Documentation

> Server-rendered guides for model providers, model families, API protocols, and use cases.

## zh-CN

- [模型与 API 文档中心](https://api.tu-zi.com/docs): 这里汇总经人工核验的厂商定位、模型家族能力、协议语义、限制和官方来源；实时型号与价格独立放在模型广场。
- [Anthropic Messages API 调用指南](https://api.tu-zi.com/docs/api/anthropic): Messages API 使用 system、角色消息和内容块；直连 Anthropic 要求 `x-api-key` 与版本头，本站推荐统一 Bearer，同时兼容原生 `x-api-key`。
- [音频、转写与 Realtime API](https://api.tu-zi.com/docs/api/audio-realtime): TTS 使用 /v1/audio/speech，文件转写与翻译使用 multipart 音频端点，OpenAI Realtime 使用 /v1/realtime；Gemini Live 使用 /v1beta/live?model=gemini-3.8-live 原生 WebSocket。SSE、二进制 HTTP 和不同 WebSocket 协议不能混为一谈。
- [Gemini API 调用指南：GenerateContent、Live 与 Interactions](https://api.tu-zi.com/docs/api/gemini): 本站提供 GenerateContent 与 gemini-3.8-live 的 /v1beta/live 原生 WebSocket；Live 使用 setup 与 serverContent 协议，只支持语音输出，可开启配套转写。
- [图像生成与编辑 API](https://api.tu-zi.com/docs/api/images): 新图使用 POST /v1/images/generations，参考图编辑使用 POST /v1/images/edits；先确认模型支持对应端点，再处理 URL/Base64 结果、超时和内容安全。
- [Kling 风格异步媒体 API](https://api.tu-zi.com/docs/api/kling): 本站提供带 /kling 前缀的 Kling 风格兼容入口；它使用本站令牌并返回本站公开 task_id，不等同于直连 api.klingai.com。提交后轮询到 SUCCESS 或 FAILURE。
- [Midjourney 任务兼容协议](https://api.tu-zi.com/docs/api/midjourney): 本站 /mj/* 是任务代理兼容协议，不是 Midjourney 官方公开 API；先确认渠道授权与条款，再提交 imagine，保存 result 任务 ID，并轮询到 SUCCESS 或 FAILURE。
- [内容 Moderation 安全分类 API](https://api.tu-zi.com/docs/api/moderation): POST /v1/moderations 返回 flagged、categories 与 category_scores；它是风险信号，不是完整的业务政策或人工审核替代品。
- [OpenAI 兼容 API 调用指南](https://api.tu-zi.com/docs/api/openai): 本站 OpenAI 兼容接口使用 Bearer 令牌；Responses、Chat Completions 与媒体端点拥有不同对象和流式语义，需按目标模型选择。
- [Embedding 与 Rerank 检索 API](https://api.tu-zi.com/docs/api/retrieval): POST /v1/embeddings 将文本转换为向量，POST /v1/rerank 对候选文档按查询重排；两者模型、输入上限和分数不可混用。
- [Seedance 真人素材视频生成完整教程](https://api.tu-zi.com/docs/api/seedance-real-person-video): 使用本站令牌上传图片到 /v1/seedance/assets，主动发起 compliance 审核；等待每张图片的 status 变为 active，再把 reference 放入 /v1/videos 的 content，保存视频任务 ID 并查询到 completed 后下载。
- [Suno 音乐任务代理协议](https://api.tu-zi.com/docs/api/suno): 本站 /suno/* 是任务代理协议，不是 Suno 官方公开 API；常用入口为 POST /suno/submit/music，返回任务 ID 后用 GET /suno/fetch/{task_id} 查询到 SUCCESS 或 FAILURE。
- [工具调用与结构化输出 API](https://api.tu-zi.com/docs/api/tools-structured): tools 只让模型提出结构化调用，应用仍需鉴权、校验和执行；strict 结构化输出约束模型响应，但业务数据仍必须再次验证。
- [视频生成任务 API](https://api.tu-zi.com/docs/api/video): POST /v1/videos 创建任务，GET /v1/videos/{task_id} 查询，completed 后再读取内容；视频是异步任务，不能把提交成功当作生成成功。
- [Anthropic Claude](https://api.tu-zi.com/docs/models/claude): 截至 2026-08-08，Anthropic 官方以 Fable 5、Opus 5、Sonnet 5 和 Haiku 4.5 分别覆盖最高能力、复杂 Agent、平衡型与低延迟任务。
- [DeepSeek](https://api.tu-zi.com/docs/models/deepseek): 当前官方 ID 是 deepseek-v4-flash 和 deepseek-v4-pro，两者默认启用思考并可关闭；本站仅保证模型列表和协议页明确启用的名称与路径。
- [豆包 Doubao](https://api.tu-zi.com/docs/models/doubao): 豆包模型在火山方舟中覆盖文本、推理、Coding、Agent 与多模态；预置模型通常在 `model` 中填写 Model ID，自定义部署才使用 Endpoint ID。
- [FLUX 图像模型](https://api.tu-zi.com/docs/models/flux): FLUX.2 官方按 klein、max、pro、flex 区分本地高速、最高一致性、生产级质量和文字排版能力，并提供不同 API 与开源许可证。
- [Google Gemini](https://api.tu-zi.com/docs/models/gemini): Gemini 是 Google 的多模态家族，官方目录按通用推理、低延迟、图像、视频、音频、Embedding 与 Agent 等任务列出独立型号和方法。
- [智谱 GLM](https://api.tu-zi.com/docs/models/glm): 智谱官方当前推荐 GLM-5.2 处理长程文本与 Coding、GLM-5V-Turbo 处理视觉 Coding、GLM-Image 处理图像生成，并提供多类专用模型。
- [Jev-1.13](https://api.tu-zi.com/docs/models/jev-1-13): 通过本站模型广场使用 /v1/systemone 提交 state 与 questions；响应从原生 answers 或 Community 的 data.answers 读取，适合分级、路由和风险判断。
- [Moonshot Kimi](https://api.tu-zi.com/docs/models/kimi): Kimi 官方当前以 K3、K2.7 Code 与 K2.6 覆盖通用多模态、长程编码和既有任务，并提供 OpenAI 兼容对话、工具与 JSON 能力。
- [Kling 可灵视频模型](https://api.tu-zi.com/docs/models/kling): Kling 是可灵开放平台的图像与视频生成家族，采用异步任务接口；当前型号、时长、分辨率和参考控制必须从动态官方文档核验。
- [Meta Llama](https://api.tu-zi.com/docs/models/llama): Llama 是 Meta 的开放权重家族，通常通过自托管或合作云调用；模型卡、许可证、提示模板和实际推理服务共同决定能力与限制。
- [Mistral 系列](https://api.tu-zi.com/docs/models/mistral): Mistral 官方当前以 Medium 3.5、Small 4 为通用/Agent 主线，并为代码、OCR、语音、Embedding 与安全提供独立型号及不同许可证。
- [OpenAI GPT 与 o 系列](https://api.tu-zi.com/docs/models/openai-gpt): 截至 2026-08-08，OpenAI 官方以 GPT-5.6 Sol、Terra、Luna 分别覆盖复杂专业任务、能力成本平衡和高吞吐场景，并通过 Responses API 提供。
- [Qwen 通义千问](https://api.tu-zi.com/docs/models/qwen): Qwen 家族覆盖通用、推理、代码、视觉语言与媒体生成；百炼官方目录按地域和协议给出精确 ID，不能把第三方模型算作 Qwen。
- [360 AI Model Platform](https://api.tu-zi.com/docs/providers/ai360-platform): 360 的模型开放平台，目录同时包含平台自有与第三方模型，归属应以单个模型说明为准。
- [360 AI Model Platform 语音与音频](https://api.tu-zi.com/docs/providers/ai360-platform/audio): 先用本文理解 360 AI Model Platform 的语音与音频能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [360 AI Model Platform 图像生成与编辑](https://api.tu-zi.com/docs/providers/ai360-platform/image): 先用本文理解 360 AI Model Platform 的图像生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [360 AI Model Platform 向量、检索与重排](https://api.tu-zi.com/docs/providers/ai360-platform/retrieval): 先用本文理解 360 AI Model Platform 的向量、检索与重排能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [360 AI Model Platform 文本、推理与代码](https://api.tu-zi.com/docs/providers/ai360-platform/text): 先用本文理解 360 AI Model Platform 的文本、推理与代码能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [360 AI Model Platform 视频生成与编辑](https://api.tu-zi.com/docs/providers/ai360-platform/video): 先用本文理解 360 AI Model Platform 的视频生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Anthropic Claude](https://api.tu-zi.com/docs/providers/anthropic): 直连 Anthropic Messages API 必须发送 `x-api-key` 与 `anthropic-version`；本站推荐 Bearer，也兼容 `/v1/messages` 的原生 `x-api-key`，且流式 HTTP 200 后仍可能出现 error 事件。
- [Anthropic Claude 文本、推理与代码](https://api.tu-zi.com/docs/providers/anthropic/text): 先用本文理解 Anthropic Claude 的文本、推理与代码能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [AWS Bedrock](https://api.tu-zi.com/docs/providers/aws-bedrock): AWS 托管的多厂商基础模型平台，提供统一推理、模型发现和云服务集成。
- [AWS Bedrock 语音与音频](https://api.tu-zi.com/docs/providers/aws-bedrock/audio): 先用本文理解 AWS Bedrock 的语音与音频能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [AWS Bedrock 图像生成与编辑](https://api.tu-zi.com/docs/providers/aws-bedrock/image): 先用本文理解 AWS Bedrock 的图像生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [AWS Bedrock 向量、检索与重排](https://api.tu-zi.com/docs/providers/aws-bedrock/retrieval): 先用本文理解 AWS Bedrock 的向量、检索与重排能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [AWS Bedrock 文本、推理与代码](https://api.tu-zi.com/docs/providers/aws-bedrock/text): 先用本文理解 AWS Bedrock 的文本、推理与代码能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [AWS Bedrock 视频生成与编辑](https://api.tu-zi.com/docs/providers/aws-bedrock/video): 先用本文理解 AWS Bedrock 的视频生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Azure AI Foundry](https://api.tu-zi.com/docs/providers/azure-ai-foundry): 直连 Foundry 时请求中的 model 通常是部署名，不是目录模型 ID；生产优先使用 Microsoft Entra ID。
- [Azure AI Foundry 语音与音频](https://api.tu-zi.com/docs/providers/azure-ai-foundry/audio): 先用本文理解 Azure AI Foundry 的语音与音频能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Azure AI Foundry 图像生成与编辑](https://api.tu-zi.com/docs/providers/azure-ai-foundry/image): 先用本文理解 Azure AI Foundry 的图像生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Azure AI Foundry 向量、检索与重排](https://api.tu-zi.com/docs/providers/azure-ai-foundry/retrieval): 先用本文理解 Azure AI Foundry 的向量、检索与重排能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Azure AI Foundry 文本、推理与代码](https://api.tu-zi.com/docs/providers/azure-ai-foundry/text): 先用本文理解 Azure AI Foundry 的文本、推理与代码能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Azure AI Foundry 视频生成与编辑](https://api.tu-zi.com/docs/providers/azure-ai-foundry/video): 先用本文理解 Azure AI Foundry 的视频生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Baidu ERNIE](https://api.tu-zi.com/docs/providers/baidu-ernie): 直连千帆使用 `https://qianfan.baidubce.com/v2` 和 Bearer `bce-v3/...` Key；预置模型填写模型列表的 model 入参，自定义部署填写服务 API 名称，本站使用本站令牌。
- [Baidu ERNIE 语音与音频](https://api.tu-zi.com/docs/providers/baidu-ernie/audio): 先用本文理解 Baidu ERNIE 的语音与音频能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Baidu ERNIE 图像生成与编辑](https://api.tu-zi.com/docs/providers/baidu-ernie/image): 先用本文理解 Baidu ERNIE 的图像生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Baidu ERNIE 向量、检索与重排](https://api.tu-zi.com/docs/providers/baidu-ernie/retrieval): 先用本文理解 Baidu ERNIE 的向量、检索与重排能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Baidu ERNIE 文本、推理与代码](https://api.tu-zi.com/docs/providers/baidu-ernie/text): 先用本文理解 Baidu ERNIE 的文本、推理与代码能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Baidu ERNIE 视频生成与编辑](https://api.tu-zi.com/docs/providers/baidu-ernie/video): 先用本文理解 Baidu ERNIE 的视频生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Black Forest Labs](https://api.tu-zi.com/docs/providers/black-forest-labs): 专注于高质量图像生成与编辑，覆盖实时预览、生产工作流和精细控制。
- [Black Forest Labs 图像生成与编辑](https://api.tu-zi.com/docs/providers/black-forest-labs/image): 先用本文理解 Black Forest Labs 的图像生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Cloudflare Workers AI](https://api.tu-zi.com/docs/providers/cloudflare-workers-ai): Cloudflare 边缘网络上的托管推理平台，提供多任务模型目录和 OpenAI 兼容接口。
- [Cloudflare Workers AI 语音与音频](https://api.tu-zi.com/docs/providers/cloudflare-workers-ai/audio): 先用本文理解 Cloudflare Workers AI 的语音与音频能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Cloudflare Workers AI 图像生成与编辑](https://api.tu-zi.com/docs/providers/cloudflare-workers-ai/image): 先用本文理解 Cloudflare Workers AI 的图像生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Cloudflare Workers AI 向量、检索与重排](https://api.tu-zi.com/docs/providers/cloudflare-workers-ai/retrieval): 先用本文理解 Cloudflare Workers AI 的向量、检索与重排能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Cloudflare Workers AI 文本、推理与代码](https://api.tu-zi.com/docs/providers/cloudflare-workers-ai/text): 先用本文理解 Cloudflare Workers AI 的文本、推理与代码能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Cohere](https://api.tu-zi.com/docs/providers/cohere): Cohere 新直连项目使用原生 `POST /v2/chat`；本站 `/v1/chat/completions` 是已适配渠道的 OpenAI 兼容入口，并非 Cohere 原生路径。Embed 必须正确选择 `input_type` 与维度，Rerank、Trial/Production 限流和模型退役也需分别核验。
- [Cohere 语音与音频](https://api.tu-zi.com/docs/providers/cohere/audio): 先用本文理解 Cohere 的语音与音频能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Cohere 向量、检索与重排](https://api.tu-zi.com/docs/providers/cohere/retrieval): 先用本文理解 Cohere 的向量、检索与重排能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Cohere 文本、推理与代码](https://api.tu-zi.com/docs/providers/cohere/text): 先用本文理解 Cohere 的文本、推理与代码能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [DeepSeek](https://api.tu-zi.com/docs/providers/deepseek): DeepSeek 当前官方模型为 deepseek-v4-flash 与 deepseek-v4-pro，默认启用思考，也可关闭；直连使用官方 API Key，本站网关使用本站 Bearer 令牌。
- [DeepSeek 文本、推理与代码](https://api.tu-zi.com/docs/providers/deepseek/text): 先用本文理解 DeepSeek 的文本、推理与代码能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [ByteDance Doubao](https://api.tu-zi.com/docs/providers/doubao): 直连火山方舟使用项目内 API Key 和 `https://ark.cn-beijing.volces.com/api/v3`；预置模型通常填写 Model ID，自定义部署才使用 Endpoint ID，本站则使用本站令牌与已启用模型名。
- [ByteDance Doubao 图像生成与编辑](https://api.tu-zi.com/docs/providers/doubao/image): 先用本文理解 ByteDance Doubao 的图像生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [ByteDance Doubao 向量、检索与重排](https://api.tu-zi.com/docs/providers/doubao/retrieval): 先用本文理解 ByteDance Doubao 的向量、检索与重排能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [ByteDance Doubao 文本、推理与代码](https://api.tu-zi.com/docs/providers/doubao/text): 先用本文理解 ByteDance Doubao 的文本、推理与代码能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [ByteDance Doubao 视频生成与编辑](https://api.tu-zi.com/docs/providers/doubao/video): 先用本文理解 ByteDance Doubao 的视频生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Google Gemini](https://api.tu-zi.com/docs/providers/google-gemini): Google 官方推荐直连新项目优先使用 Interactions API；本站当前 Gemini 原生路径确定为 GenerateContent 与 streamGenerateContent，不提供 Interactions，认证推荐 Bearer 并兼容 x-goog-api-key。
- [Google Gemini 语音与音频](https://api.tu-zi.com/docs/providers/google-gemini/audio): 先用本文理解 Google Gemini 的语音与音频能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Google Gemini 图像生成与编辑](https://api.tu-zi.com/docs/providers/google-gemini/image): 先用本文理解 Google Gemini 的图像生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Google Gemini 向量、检索与重排](https://api.tu-zi.com/docs/providers/google-gemini/retrieval): 先用本文理解 Google Gemini 的向量、检索与重排能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Google Gemini 文本、推理与代码](https://api.tu-zi.com/docs/providers/google-gemini/text): 先用本文理解 Google Gemini 的文本、推理与代码能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Google Gemini 视频生成与编辑](https://api.tu-zi.com/docs/providers/google-gemini/video): 先用本文理解 Google Gemini 的视频生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [HappyHorse](https://api.tu-zi.com/docs/providers/happyhorse): 直连使用阿里云地域/Workspace 对应的 API Key、Endpoint 和强制异步头；本站 `/v1/videos` 使用本站令牌。任务 ID 与视频 URL 仅保留 24 小时，应及时查询和转存。
- [HappyHorse 视频生成与编辑](https://api.tu-zi.com/docs/providers/happyhorse/video): 先用本文理解 HappyHorse 的视频生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [iFLYTEK Spark](https://api.tu-zi.com/docs/providers/iflytek-spark): 讯飞星火的通用 HTTP API、Spark-X1.5 与 Spark-X2 使用不同 Base URL；官方直连使用对应服务的 APIPassword，本站统一使用本站 Bearer 令牌和已适配路径。
- [iFLYTEK Spark 文本、推理与代码](https://api.tu-zi.com/docs/providers/iflytek-spark/text): 先用本文理解 iFLYTEK Spark 的文本、推理与代码能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [JevAI 结构化决策模型](https://api.tu-zi.com/docs/providers/jevai): JEV 适合把业务事实转换为 choice、score 和 noul 决策信号；在本站使用 JEV Community 时，请通过模型广场的 /v1/systemone 端点调用。
- [JevAI 结构化决策模型 文本、推理与代码](https://api.tu-zi.com/docs/providers/jevai/text): 先用本文理解 JevAI 结构化决策模型 的文本、推理与代码能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [ByteDance Jimeng](https://api.tu-zi.com/docs/providers/jimeng): 直连即梦使用 `visual.volcengineapi.com` 与火山引擎签名，本站 `/v1/videos` 使用本站 Bearer 令牌；客户端必须保存 task_id、处理终态并及时转存临时媒体。
- [ByteDance Jimeng 图像生成与编辑](https://api.tu-zi.com/docs/providers/jimeng/image): 先用本文理解 ByteDance Jimeng 的图像生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [ByteDance Jimeng 视频生成与编辑](https://api.tu-zi.com/docs/providers/jimeng/video): 先用本文理解 ByteDance Jimeng 的视频生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Jina AI](https://api.tu-zi.com/docs/providers/jina-ai): Jina AI 既是搜索基础模型厂商，也是托管 API 提供方，不是通用聊天模型目录；直连使用 Jina Bearer Key，本站仅以本站密钥开放已适配的 Embeddings 与 Rerank 入口。
- [Jina AI 向量、检索与重排](https://api.tu-zi.com/docs/providers/jina-ai/retrieval): 先用本文理解 Jina AI 的向量、检索与重排能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Kuaishou Kling AI](https://api.tu-zi.com/docs/providers/kling): 快手可灵开放平台提供面向创意内容生产的图像与视频生成接口。
- [Kuaishou Kling AI 图像生成与编辑](https://api.tu-zi.com/docs/providers/kling/image): 先用本文理解 Kuaishou Kling AI 的图像生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Kuaishou Kling AI 视频生成与编辑](https://api.tu-zi.com/docs/providers/kling/video): 先用本文理解 Kuaishou Kling AI 的视频生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Meta Llama](https://api.tu-zi.com/docs/providers/meta-llama): Meta 的开放权重模型家族，可自行部署或通过合作云平台使用。
- [Meta Llama 文本、推理与代码](https://api.tu-zi.com/docs/providers/meta-llama/text): 先用本文理解 Meta Llama 的文本、推理与代码能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [MiniMax](https://api.tu-zi.com/docs/providers/minimax): 中国站直连 OpenAI Base URL 为 `https://api.minimaxi.com/v1`，Anthropic 兼容为 `/anthropic`；按量与 Token Plan 使用独立 Key，本站仅使用本站令牌和适配路径。
- [MiniMax 语音与音频](https://api.tu-zi.com/docs/providers/minimax/audio): 先用本文理解 MiniMax 的语音与音频能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [MiniMax 图像生成与编辑](https://api.tu-zi.com/docs/providers/minimax/image): 先用本文理解 MiniMax 的图像生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [MiniMax 文本、推理与代码](https://api.tu-zi.com/docs/providers/minimax/text): 先用本文理解 MiniMax 的文本、推理与代码能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [MiniMax 视频生成与编辑](https://api.tu-zi.com/docs/providers/minimax/video): 先用本文理解 MiniMax 的视频生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Mistral AI](https://api.tu-zi.com/docs/providers/mistral): 直连 Mistral 首调使用 Mistral Base URL、`Authorization: Bearer` 和精确模型 ID；本站只保留已适配的 `/v1/chat/completions`，必须改用本站 Base URL 与密钥。Batch、Guardrails 和 Moderation 是 Mistral 官方能力，不代表本站入口自动支持或自动开启。
- [Mistral AI 语音与音频](https://api.tu-zi.com/docs/providers/mistral/audio): 先用本文理解 Mistral AI 的语音与音频能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Mistral AI 向量、检索与重排](https://api.tu-zi.com/docs/providers/mistral/retrieval): 先用本文理解 Mistral AI 的向量、检索与重排能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Mistral AI 文本、推理与代码](https://api.tu-zi.com/docs/providers/mistral/text): 先用本文理解 Mistral AI 的文本、推理与代码能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Moonshot Kimi](https://api.tu-zi.com/docs/providers/moonshot-kimi): 中国站直连使用 `https://api.moonshot.cn/v1` 和 Moonshot API Key；K3、K2.6、K2.7 的思考与工具参数不同，本站则使用本站令牌和已启用模型名。
- [Moonshot Kimi 文本、推理与代码](https://api.tu-zi.com/docs/providers/moonshot-kimi/text): 先用本文理解 Moonshot Kimi 的文本、推理与代码能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Ollama](https://api.tu-zi.com/docs/providers/ollama): Ollama 同时提供本地运行时与云端 API；两者 Base URL 和认证方式不同。
- [Ollama 向量、检索与重排](https://api.tu-zi.com/docs/providers/ollama/retrieval): 先用本文理解 Ollama 的向量、检索与重排能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Ollama 文本、推理与代码](https://api.tu-zi.com/docs/providers/ollama/text): 先用本文理解 Ollama 的文本、推理与代码能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [OpenAI](https://api.tu-zi.com/docs/providers/openai): OpenAI 官方推荐新项目优先使用 Responses API；直连和本站都使用 Bearer 请求头，但必须分别使用 OpenAI 密钥与本站密钥，并确认本站已适配目标端点。
- [OpenAI 语音与音频](https://api.tu-zi.com/docs/providers/openai/audio): 先用本文理解 OpenAI 的语音与音频能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [OpenAI 图像生成与编辑](https://api.tu-zi.com/docs/providers/openai/image): 先用本文理解 OpenAI 的图像生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [OpenAI 向量、检索与重排](https://api.tu-zi.com/docs/providers/openai/retrieval): 先用本文理解 OpenAI 的向量、检索与重排能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [OpenAI 文本、推理与代码](https://api.tu-zi.com/docs/providers/openai/text): 先用本文理解 OpenAI 的文本、推理与代码能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [OpenAI 视频生成与编辑](https://api.tu-zi.com/docs/providers/openai/video): 先用本文理解 OpenAI 的视频生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Alibaba Qwen](https://api.tu-zi.com/docs/providers/qwen): 直连百炼应复制目标地域与 Workspace 的 API Host，并使用同一范围的 Key；新版按量 Key 常以 sk-ws- 开头，套餐 Key 以 sk-sp- 开头，本站则使用本站令牌。
- [Alibaba Qwen 语音与音频](https://api.tu-zi.com/docs/providers/qwen/audio): 先用本文理解 Alibaba Qwen 的语音与音频能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Alibaba Qwen 图像生成与编辑](https://api.tu-zi.com/docs/providers/qwen/image): 先用本文理解 Alibaba Qwen 的图像生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Alibaba Qwen 向量、检索与重排](https://api.tu-zi.com/docs/providers/qwen/retrieval): 先用本文理解 Alibaba Qwen 的向量、检索与重排能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Alibaba Qwen 文本、推理与代码](https://api.tu-zi.com/docs/providers/qwen/text): 先用本文理解 Alibaba Qwen 的文本、推理与代码能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Alibaba Qwen 视频生成与编辑](https://api.tu-zi.com/docs/providers/qwen/video): 先用本文理解 Alibaba Qwen 的视频生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Tencent Hunyuan](https://api.tu-zi.com/docs/providers/tencent-hunyuan): 直连 TokenHub 应先调用 `/v1/models` 获取区分大小写的模型 ID；自定义在线推理才使用服务 ID。广州与国际站地址不同，本站使用本站令牌和适配路径。
- [Tencent Hunyuan 图像生成与编辑](https://api.tu-zi.com/docs/providers/tencent-hunyuan/image): 先用本文理解 Tencent Hunyuan 的图像生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Tencent Hunyuan 向量、检索与重排](https://api.tu-zi.com/docs/providers/tencent-hunyuan/retrieval): 先用本文理解 Tencent Hunyuan 的向量、检索与重排能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Tencent Hunyuan 文本、推理与代码](https://api.tu-zi.com/docs/providers/tencent-hunyuan/text): 先用本文理解 Tencent Hunyuan 的文本、推理与代码能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Tencent Hunyuan 视频生成与编辑](https://api.tu-zi.com/docs/providers/tencent-hunyuan/video): 先用本文理解 Tencent Hunyuan 的视频生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Vidu](https://api.tu-zi.com/docs/providers/vidu): 提供文本、图像和参考素材驱动的视频生成，并覆盖图像与音效能力。
- [Vidu 图像生成与编辑](https://api.tu-zi.com/docs/providers/vidu/image): 先用本文理解 Vidu 的图像生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Vidu 视频生成与编辑](https://api.tu-zi.com/docs/providers/vidu/video): 先用本文理解 Vidu 的视频生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [xAI Grok](https://api.tu-zi.com/docs/providers/xai): 面向通用生成、推理、编码、工具调用以及实时语音和媒体生成的模型平台。
- [xAI Grok 图像生成与编辑](https://api.tu-zi.com/docs/providers/xai/image): 先用本文理解 xAI Grok 的图像生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [xAI Grok 文本、推理与代码](https://api.tu-zi.com/docs/providers/xai/text): 先用本文理解 xAI Grok 的文本、推理与代码能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [xAI Grok 视频生成与编辑](https://api.tu-zi.com/docs/providers/xai/video): 先用本文理解 xAI Grok 的视频生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [01.AI Yi](https://api.tu-zi.com/docs/providers/yi): 直连 Yi 应使用零一万物控制台当前给出的 API Key、Base URL 和模型 ID；本站使用本站 Bearer 令牌与 `/v1/chat/completions`，两类凭证不能混用。
- [01.AI Yi 文本、推理与代码](https://api.tu-zi.com/docs/providers/yi/text): 先用本文理解 01.AI Yi 的文本、推理与代码能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Zhipu GLM](https://api.tu-zi.com/docs/providers/zhipu-glm): 直连 GLM 使用 `https://open.bigmodel.cn/api/paas/v4/`、Bearer API Key 和精确模型 ID；本站使用本站令牌，且只保证本页声明的适配路径。
- [Zhipu GLM 语音与音频](https://api.tu-zi.com/docs/providers/zhipu-glm/audio): 先用本文理解 Zhipu GLM 的语音与音频能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Zhipu GLM 图像生成与编辑](https://api.tu-zi.com/docs/providers/zhipu-glm/image): 先用本文理解 Zhipu GLM 的图像生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Zhipu GLM 向量、检索与重排](https://api.tu-zi.com/docs/providers/zhipu-glm/retrieval): 先用本文理解 Zhipu GLM 的向量、检索与重排能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Zhipu GLM 文本、推理与代码](https://api.tu-zi.com/docs/providers/zhipu-glm/text): 先用本文理解 Zhipu GLM 的文本、推理与代码能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [Zhipu GLM 视频生成与编辑](https://api.tu-zi.com/docs/providers/zhipu-glm/video): 先用本文理解 Zhipu GLM 的视频生成与编辑能力，再到模型广场核对精确可调用 ID、端点与价格后接入。
- [工具调用与 AI Agent 模型指南](https://api.tu-zi.com/docs/use-cases/agents): Agent 的核心不是会调用工具，而是受控循环：结构化 schema、最小权限、状态、幂等、超时、人工确认、审计和可回放评测。
- [统一异步任务：提交、轮询与结果生命周期](https://api.tu-zi.com/docs/use-cases/async-tasks): 对适合后台执行的 POST 请求，可在原路径前加 `/async` 提交并用 `/get-async?id=...` 查询；流式、模型列表、Files 和 Realtime 不能这样包装。
- [语音、转写与实时音频 API 指南](https://api.tu-zi.com/docs/use-cases/audio): 转写、语音合成和实时语音是不同链路，分别关注识别准确率、音色授权与低延迟双工传输，不能用一个音频端点概括。
- [AI 模型的 API 聚合平台新手指南：从注册到调用与查账](https://api.tu-zi.com/docs/use-cases/beginner-guide): 新手按“注册登录 → 模型广场确认精确模型 ID、端点和价格 → 充值 → 创建有限额令牌 → 在线测试或最小请求 → 按 Request-ID 查日志”的顺序操作，最容易定位问题并控制费用。
- [计费、用量与成本核对](https://api.tu-zi.com/docs/use-cases/billing-usage): 价格以模型广场和当前分组为准；请求侧保存模型、路径、Request-ID、usage 或任务 ID，再用消费日志核对最终结算，不能只看提交时余额变化。
- [模型原厂、托管云、聚合渠道与本地运行时](https://api.tu-zi.com/docs/use-cases/channel-taxonomy): 厂商页回答模型是谁开发的，渠道回答请求从哪里转发；同名模型经过不同托管或聚合渠道时，认证、协议、区域、版本和数据边界都可能变化。
- [AI 编码模型选择与调用](https://api.tu-zi.com/docs/use-cases/coding): 编码模型选型应以真实仓库任务的补丁正确率、测试通过率、工具成功率、延迟和成本为依据，而不是模型广场描述或单项榜单。
- [豆包计费：在线、Fast、批量与缓存](https://api.tu-zi.com/docs/use-cases/doubao-billing): 在线适合即时交互，Fast 是受模型与渠道支持约束的低延迟服务档位，批量是独立的上游推理服务；普通并发不会自动享受批量价。费用还要结合输入长度、音频、缓存和当前分组核对。
- [本站 API 端点与能力矩阵](https://api.tu-zi.com/docs/use-cases/endpoint-matrix): 先在模型广场确认精确模型与已开放端点，再按本页选择协议；路由存在只表示网关认识该路径，不表示每个模型或渠道都支持。
- [模型评测、灰度与版本迁移](https://api.tu-zi.com/docs/use-cases/evaluation-migration): 不要以排行榜或单个提示词决定升级；先固定当前模型、参数和评测集，再比较候选版本，灰度观察质量、延迟、错误与成本后才能切换。
- [文件与媒体输入、输出及生命周期](https://api.tu-zi.com/docs/use-cases/files-media): 本站 `/v1/files` 当前未实现；文件必须按目标模型和端点支持的方式随请求提交，所有输入与结果 URL 都应视为有权限和生命周期的外部资源。
- [AI 网关管理员手册：渠道上线、重试与供应商健康](https://api.tu-zi.com/docs/use-cases/gateway-operations): 新渠道应先保持隔离，核对类型、Base URL、密钥、模型与映射后执行单渠道测试，再小流量启用；重试必须受次数、时间和幂等边界约束，健康面板与供应商账户用于判断是协议、渠道还是余额故障。
- [AI API 快速入门：从 API Key 到第一次调用](https://api.tu-zi.com/docs/use-cases/getting-started): 调用 AI API 需要匹配的 Base URL、API Key、模型名和端点；先用最小 curl 请求跑通，再处理错误、费用与密钥安全。
- [图像生成与视觉模型指南](https://api.tu-zi.com/docs/use-cases/image-generation): 视觉理解与图像生成是两类接口；生成项目要分别验证提示遵循、编辑一致性、文字、人物权利、安全审核和输出许可。
- [媒体与对象存储管理员手册：Profile、权限和生命周期](https://api.tu-zi.com/docs/use-cases/media-storage-operations): 按用途绑定独立存储 Profile，先测试连接再保存并同步生命周期；生成与中转结果可按期限清理，合规和消息资产应使用私有代理与独立保留策略，签名 URL 与公开 URL 都按凭证管理。
- [AI API 生产可靠性：超时、重试、限流与 SLO](https://api.tu-zi.com/docs/use-cases/production-reliability): 生产可靠性依赖分阶段超时、有抖动的有限重试、并发与速率双重保护、幂等边界、熔断降级和以完成事件为准的 SLO；不能靠延长总超时或无限重试解决。
- [RAG、Embedding 与重排模型指南](https://api.tu-zi.com/docs/use-cases/rag): RAG 是切分、Embedding、召回、重排、证据生成和评测组成的系统；完整性来自可引用证据和无答案处理，不来自更大的模型目录。
- [AI API 安全、隐私与数据治理](https://api.tu-zi.com/docs/use-cases/security-governance): 把模型视为不可信内容处理器：密钥只在服务端保存，输入先分级，外部 URL 受限抓取，工具参数重新授权，日志脱敏，并核对每个实际上游的数据政策。
- [SSE 流式响应与 WebSocket 实时连接](https://api.tu-zi.com/docs/use-cases/streaming): SSE 是单次 HTTP 响应的事件流，WebSocket 是双向长连接；客户端必须按协议事件判断完成，不能把 HTTP 200 或连接关闭当成业务成功。
- [通用文本、对话与推理模型指南](https://api.tu-zi.com/docs/use-cases/text-reasoning): 通用文本模型覆盖问答、写作、抽取和分析，推理模型更适合多步问题；生产选型应以真实任务正确率、延迟、成本和可控性共同判断。
- [AI API 通用错误码与排障指南](https://api.tu-zi.com/docs/use-cases/troubleshooting): 处理 AI API 错误时应同时解析 HTTP 状态、厂商业务码、错误类型、消息、重试提示和请求 ID；参数、鉴权、余额与硬配额错误先修复根因，只有临时限流、网络错误和确认属于瞬时故障的部分 5xx 才适合有限重试。
- [视频生成模型与异步 API 指南](https://api.tu-zi.com/docs/use-cases/video-generation): 视频生成应按异步任务系统设计：幂等提交、持久化任务 ID、有界轮询或验签回调、终态处理、结果转存与内容审核缺一不可。

## en

- [Model and API documentation](https://api.tu-zi.com/en/docs): This center contains human-reviewed provider positioning, model-family capabilities, protocol semantics, limits, and official sources; live IDs and prices remain in the marketplace.
- [Anthropic Messages API guide](https://api.tu-zi.com/en/docs/api/anthropic): Messages uses system instructions, role messages, and content blocks; direct Anthropic calls require `x-api-key` and a version header, while this site recommends Bearer and also accepts native `x-api-key`.
- [Audio, transcription, and Realtime API](https://api.tu-zi.com/en/docs/api/audio-realtime): Use /v1/audio/speech for TTS, multipart audio endpoints for file transcription or translation, /v1/realtime for OpenAI Realtime, and /v1beta/live?model=gemini-3.8-live for native Gemini Live WebSocket sessions.
- [Gemini API guide: GenerateContent, Live, and Interactions](https://api.tu-zi.com/en/docs/api/gemini): The gateway exposes GenerateContent and the native /v1beta/live WebSocket for gemini-3.8-live. Live uses setup and serverContent, with audio output and optional transcription.
- [Image generation and editing API](https://api.tu-zi.com/en/docs/api/images): Use POST /v1/images/generations for new images and POST /v1/images/edits for reference-image editing. Confirm endpoint support per model before processing URL or Base64 results.
- [Kling-style asynchronous media API](https://api.tu-zi.com/en/docs/api/kling): This site exposes Kling-style routes under /kling. They use a site token and return a public site task_id rather than acting as an unconditional pass-through to api.klingai.com. Poll to SUCCESS or FAILURE.
- [Midjourney task compatibility protocol](https://api.tu-zi.com/en/docs/api/midjourney): The /mj/* routes are this site's task-proxy compatibility contract, not an official Midjourney public API. Confirm channel authorization and current terms, persist result as the task ID, and poll to SUCCESS or FAILURE.
- [Content Moderation safety classification API](https://api.tu-zi.com/en/docs/api/moderation): POST /v1/moderations returns flagged, categories, and category_scores. Treat them as risk signals, not a complete business policy or substitute for human review.
- [OpenAI-compatible API guide](https://api.tu-zi.com/en/docs/api/openai): This site's OpenAI-compatible routes use a Bearer token; Responses, Chat Completions, and media endpoints have distinct objects and streaming semantics and must match the target model.
- [Embedding and Rerank retrieval API](https://api.tu-zi.com/en/docs/api/retrieval): POST /v1/embeddings maps text to vectors; POST /v1/rerank orders candidate documents for a query. Their models, limits, and scores are not interchangeable.
- [Complete Guide to Seedance Video Generation with Real-Person Images](https://api.tu-zi.com/en/docs/api/seedance-real-person-video): Upload images to /v1/seedance/assets using your site API token, explicitly start a compliance review, and wait for every image to become active. Add each reference to the content array of /v1/videos, save the video task ID, and download the result after its status becomes completed.
- [Suno music task-proxy protocol](https://api.tu-zi.com/en/docs/api/suno): The /suno/* routes are this site's task-proxy protocol, not an official Suno public API. POST /suno/submit/music returns a task ID for GET /suno/fetch/{task_id}; poll to SUCCESS or FAILURE.
- [Tool calling and structured output API](https://api.tu-zi.com/en/docs/api/tools-structured): tools let a model propose a structured call; the application must still authorize, validate, execute, and return the result. strict constrains shape, not business truth.
- [Video generation task API](https://api.tu-zi.com/en/docs/api/video): Create with POST /v1/videos, poll GET /v1/videos/{task_id}, and fetch content only after completed. Submission success is not generation success.
- [Anthropic Claude](https://api.tu-zi.com/en/docs/models/claude): As reviewed on 2026-08-08, Anthropic positions Fable 5, Opus 5, Sonnet 5, and Haiku 4.5 for highest capability, complex agents, balanced work, and low latency.
- [DeepSeek](https://api.tu-zi.com/en/docs/models/deepseek): The current official IDs are deepseek-v4-flash and deepseek-v4-pro; both enable thinking by default and can disable it. This gateway supports only names and paths explicitly enabled in its model list and protocol guide.
- [Doubao](https://api.tu-zi.com/en/docs/models/doubao): Doubao models in Volcengine Ark cover text, reasoning, coding, agents, and multimodality. Prebuilt models normally use a Model ID in `model`; custom deployments use an Endpoint ID.
- [FLUX image models](https://api.tu-zi.com/en/docs/models/flux): FLUX.2 officially separates klein, max, pro, and flex for fast local inference, maximum consistency, production quality, and typography, with different APIs and licenses.
- [Google Gemini](https://api.tu-zi.com/en/docs/models/gemini): Gemini is Google's multimodal family; the official catalog lists separate models and methods for general reasoning, low latency, images, video, audio, embeddings, and agents.
- [Zhipu GLM](https://api.tu-zi.com/en/docs/models/glm): Zhipu currently recommends GLM-5.2 for long-running text and coding, GLM-5V-Turbo for visual coding, and GLM-Image for image generation, alongside specialized models.
- [Jev-1.13](https://api.tu-zi.com/en/docs/models/jev-1-13): Call /v1/systemone from the marketplace with state and questions, then read native answers or Community data.answers for triage, routing, and risk decisions.
- [Moonshot Kimi](https://api.tu-zi.com/en/docs/models/kimi): Kimi currently documents K3, K2.7 Code, and K2.6 for general multimodal work, long-running coding, and existing workloads through an OpenAI-compatible API.
- [Kling video models](https://api.tu-zi.com/en/docs/models/kling): Kling is the Kling Open Platform image and video family with asynchronous jobs; current models, duration, resolution, and reference controls must be verified in the live official documentation.
- [Meta Llama](https://api.tu-zi.com/en/docs/models/llama): Llama is Meta's open-weight family, normally used through self-hosting or cloud partners; model cards, licenses, prompt templates, and the actual runtime jointly define behavior.
- [Mistral family](https://api.tu-zi.com/en/docs/models/mistral): Mistral currently features Medium 3.5 and Small 4 for general and agentic work, plus dedicated code, OCR, speech, embedding, and safety models with distinct licenses.
- [OpenAI GPT and o-series](https://api.tu-zi.com/en/docs/models/openai-gpt): As reviewed on 2026-08-08, OpenAI positions GPT-5.6 Sol, Terra, and Luna for complex professional work, balanced intelligence and cost, and high-volume workloads through the Responses API.
- [Qwen](https://api.tu-zi.com/en/docs/models/qwen): Qwen spans general, reasoning, coding, vision-language, and media generation; Model Studio publishes exact regional and protocol IDs, while third-party listings remain separate.
- [360 AI Model Platform](https://api.tu-zi.com/en/docs/providers/ai360-platform): The 360 model platform includes both first-party and third-party models; attribution should follow each model's documentation.
- [360 AI Model Platform Speech and audio](https://api.tu-zi.com/en/docs/providers/ai360-platform/audio): Use this guide to understand the 360 AI Model Platform Speech and audio capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [360 AI Model Platform Image generation and editing](https://api.tu-zi.com/en/docs/providers/ai360-platform/image): Use this guide to understand the 360 AI Model Platform Image generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [360 AI Model Platform Embeddings, retrieval, and reranking](https://api.tu-zi.com/en/docs/providers/ai360-platform/retrieval): Use this guide to understand the 360 AI Model Platform Embeddings, retrieval, and reranking capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [360 AI Model Platform Text, reasoning, and code](https://api.tu-zi.com/en/docs/providers/ai360-platform/text): Use this guide to understand the 360 AI Model Platform Text, reasoning, and code capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [360 AI Model Platform Video generation and editing](https://api.tu-zi.com/en/docs/providers/ai360-platform/video): Use this guide to understand the 360 AI Model Platform Video generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Anthropic Claude](https://api.tu-zi.com/en/docs/providers/anthropic): Direct Anthropic Messages calls require `x-api-key` and `anthropic-version`. This site recommends Bearer and also accepts native `x-api-key` on `/v1/messages`; an HTTP 200 stream can still carry an error event.
- [Anthropic Claude Text, reasoning, and code](https://api.tu-zi.com/en/docs/providers/anthropic/text): Use this guide to understand the Anthropic Claude Text, reasoning, and code capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [AWS Bedrock](https://api.tu-zi.com/en/docs/providers/aws-bedrock): AWS-managed platform for foundation models from multiple providers, with unified inference, model discovery, and cloud integrations.
- [AWS Bedrock Speech and audio](https://api.tu-zi.com/en/docs/providers/aws-bedrock/audio): Use this guide to understand the AWS Bedrock Speech and audio capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [AWS Bedrock Image generation and editing](https://api.tu-zi.com/en/docs/providers/aws-bedrock/image): Use this guide to understand the AWS Bedrock Image generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [AWS Bedrock Embeddings, retrieval, and reranking](https://api.tu-zi.com/en/docs/providers/aws-bedrock/retrieval): Use this guide to understand the AWS Bedrock Embeddings, retrieval, and reranking capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [AWS Bedrock Text, reasoning, and code](https://api.tu-zi.com/en/docs/providers/aws-bedrock/text): Use this guide to understand the AWS Bedrock Text, reasoning, and code capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [AWS Bedrock Video generation and editing](https://api.tu-zi.com/en/docs/providers/aws-bedrock/video): Use this guide to understand the AWS Bedrock Video generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Azure AI Foundry](https://api.tu-zi.com/en/docs/providers/azure-ai-foundry): For direct Foundry calls, model usually means the deployment name rather than the catalog model ID; prefer Microsoft Entra ID in production.
- [Azure AI Foundry Speech and audio](https://api.tu-zi.com/en/docs/providers/azure-ai-foundry/audio): Use this guide to understand the Azure AI Foundry Speech and audio capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Azure AI Foundry Image generation and editing](https://api.tu-zi.com/en/docs/providers/azure-ai-foundry/image): Use this guide to understand the Azure AI Foundry Image generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Azure AI Foundry Embeddings, retrieval, and reranking](https://api.tu-zi.com/en/docs/providers/azure-ai-foundry/retrieval): Use this guide to understand the Azure AI Foundry Embeddings, retrieval, and reranking capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Azure AI Foundry Text, reasoning, and code](https://api.tu-zi.com/en/docs/providers/azure-ai-foundry/text): Use this guide to understand the Azure AI Foundry Text, reasoning, and code capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Azure AI Foundry Video generation and editing](https://api.tu-zi.com/en/docs/providers/azure-ai-foundry/video): Use this guide to understand the Azure AI Foundry Video generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Baidu ERNIE](https://api.tu-zi.com/en/docs/providers/baidu-ernie): Direct Qianfan access uses `https://qianfan.baidubce.com/v2` and a Bearer `bce-v3/...` key. Preset models use the documented model parameter; custom deployments use a service API name, while this gateway uses its own token.
- [Baidu ERNIE Speech and audio](https://api.tu-zi.com/en/docs/providers/baidu-ernie/audio): Use this guide to understand the Baidu ERNIE Speech and audio capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Baidu ERNIE Image generation and editing](https://api.tu-zi.com/en/docs/providers/baidu-ernie/image): Use this guide to understand the Baidu ERNIE Image generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Baidu ERNIE Embeddings, retrieval, and reranking](https://api.tu-zi.com/en/docs/providers/baidu-ernie/retrieval): Use this guide to understand the Baidu ERNIE Embeddings, retrieval, and reranking capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Baidu ERNIE Text, reasoning, and code](https://api.tu-zi.com/en/docs/providers/baidu-ernie/text): Use this guide to understand the Baidu ERNIE Text, reasoning, and code capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Baidu ERNIE Video generation and editing](https://api.tu-zi.com/en/docs/providers/baidu-ernie/video): Use this guide to understand the Baidu ERNIE Video generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Black Forest Labs](https://api.tu-zi.com/en/docs/providers/black-forest-labs): Focused on high-quality image generation and editing across realtime previews, production workflows, and fine-grained control.
- [Black Forest Labs Image generation and editing](https://api.tu-zi.com/en/docs/providers/black-forest-labs/image): Use this guide to understand the Black Forest Labs Image generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Cloudflare Workers AI](https://api.tu-zi.com/en/docs/providers/cloudflare-workers-ai): A managed inference platform on Cloudflare's edge network with a multi-task model catalog and OpenAI-compatible APIs.
- [Cloudflare Workers AI Speech and audio](https://api.tu-zi.com/en/docs/providers/cloudflare-workers-ai/audio): Use this guide to understand the Cloudflare Workers AI Speech and audio capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Cloudflare Workers AI Image generation and editing](https://api.tu-zi.com/en/docs/providers/cloudflare-workers-ai/image): Use this guide to understand the Cloudflare Workers AI Image generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Cloudflare Workers AI Embeddings, retrieval, and reranking](https://api.tu-zi.com/en/docs/providers/cloudflare-workers-ai/retrieval): Use this guide to understand the Cloudflare Workers AI Embeddings, retrieval, and reranking capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Cloudflare Workers AI Text, reasoning, and code](https://api.tu-zi.com/en/docs/providers/cloudflare-workers-ai/text): Use this guide to understand the Cloudflare Workers AI Text, reasoning, and code capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Cohere](https://api.tu-zi.com/en/docs/providers/cohere): New direct Cohere integrations use native `POST /v2/chat`. This site's `/v1/chat/completions` is an OpenAI-compatible route to an adapted channel, not Cohere's native path. Embed requires the correct `input_type` and dimension, while Rerank, Trial/Production limits, and model retirement must be checked separately.
- [Cohere Speech and audio](https://api.tu-zi.com/en/docs/providers/cohere/audio): Use this guide to understand the Cohere Speech and audio capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Cohere Embeddings, retrieval, and reranking](https://api.tu-zi.com/en/docs/providers/cohere/retrieval): Use this guide to understand the Cohere Embeddings, retrieval, and reranking capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Cohere Text, reasoning, and code](https://api.tu-zi.com/en/docs/providers/cohere/text): Use this guide to understand the Cohere Text, reasoning, and code capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [DeepSeek](https://api.tu-zi.com/en/docs/providers/deepseek): DeepSeek's current official model IDs are deepseek-v4-flash and deepseek-v4-pro; thinking is enabled by default and can be disabled. Direct calls use a DeepSeek API key, while this gateway uses its own Bearer token.
- [DeepSeek Text, reasoning, and code](https://api.tu-zi.com/en/docs/providers/deepseek/text): Use this guide to understand the DeepSeek Text, reasoning, and code capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [ByteDance Doubao](https://api.tu-zi.com/en/docs/providers/doubao): Direct Ark access uses a project API key and `https://ark.cn-beijing.volces.com/api/v3`. Preset models normally use a model ID, custom deployments use an endpoint ID, and this gateway uses its own token and enabled model name.
- [ByteDance Doubao Image generation and editing](https://api.tu-zi.com/en/docs/providers/doubao/image): Use this guide to understand the ByteDance Doubao Image generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [ByteDance Doubao Embeddings, retrieval, and reranking](https://api.tu-zi.com/en/docs/providers/doubao/retrieval): Use this guide to understand the ByteDance Doubao Embeddings, retrieval, and reranking capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [ByteDance Doubao Text, reasoning, and code](https://api.tu-zi.com/en/docs/providers/doubao/text): Use this guide to understand the ByteDance Doubao Text, reasoning, and code capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [ByteDance Doubao Video generation and editing](https://api.tu-zi.com/en/docs/providers/doubao/video): Use this guide to understand the ByteDance Doubao Video generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Google Gemini](https://api.tu-zi.com/en/docs/providers/google-gemini): Google recommends Interactions API for new direct integrations. This gateway currently exposes Gemini-native GenerateContent and streamGenerateContent, not Interactions, and recommends Bearer while accepting x-goog-api-key compatibility.
- [Google Gemini Speech and audio](https://api.tu-zi.com/en/docs/providers/google-gemini/audio): Use this guide to understand the Google Gemini Speech and audio capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Google Gemini Image generation and editing](https://api.tu-zi.com/en/docs/providers/google-gemini/image): Use this guide to understand the Google Gemini Image generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Google Gemini Embeddings, retrieval, and reranking](https://api.tu-zi.com/en/docs/providers/google-gemini/retrieval): Use this guide to understand the Google Gemini Embeddings, retrieval, and reranking capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Google Gemini Text, reasoning, and code](https://api.tu-zi.com/en/docs/providers/google-gemini/text): Use this guide to understand the Google Gemini Text, reasoning, and code capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Google Gemini Video generation and editing](https://api.tu-zi.com/en/docs/providers/google-gemini/video): Use this guide to understand the Google Gemini Video generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [HappyHorse](https://api.tu-zi.com/en/docs/providers/happyhorse): Direct access uses a region- and workspace-matched Alibaba Cloud key, endpoint, and required asynchronous header; this gateway's `/v1/videos` uses its own token. Task IDs and video URLs last only 24 hours and must be retrieved promptly.
- [HappyHorse Video generation and editing](https://api.tu-zi.com/en/docs/providers/happyhorse/video): Use this guide to understand the HappyHorse Video generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [iFLYTEK Spark](https://api.tu-zi.com/en/docs/providers/iflytek-spark): Spark's general HTTP API, Spark-X1.5, and Spark-X2 use different base URLs. Direct calls use the APIPassword for that service, while this gateway uses its own Bearer token and adapted path.
- [iFLYTEK Spark Text, reasoning, and code](https://api.tu-zi.com/en/docs/providers/iflytek-spark/text): Use this guide to understand the iFLYTEK Spark Text, reasoning, and code capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [JevAI structured decision models](https://api.tu-zi.com/en/docs/providers/jevai): JEV turns business facts into choice, score, and noul signals. On this gateway, use the /v1/systemone endpoint for the Community integration.
- [JevAI structured decision models Text, reasoning, and code](https://api.tu-zi.com/en/docs/providers/jevai/text): Use this guide to understand the JevAI structured decision models Text, reasoning, and code capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [ByteDance Jimeng](https://api.tu-zi.com/en/docs/providers/jimeng): Direct Jimeng access uses `visual.volcengineapi.com` and Volcengine signing, while this gateway's `/v1/videos` uses its own Bearer token. Persist task IDs, handle terminal states, and copy temporary media promptly.
- [ByteDance Jimeng Image generation and editing](https://api.tu-zi.com/en/docs/providers/jimeng/image): Use this guide to understand the ByteDance Jimeng Image generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [ByteDance Jimeng Video generation and editing](https://api.tu-zi.com/en/docs/providers/jimeng/video): Use this guide to understand the ByteDance Jimeng Video generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Jina AI](https://api.tu-zi.com/en/docs/providers/jina-ai): Jina AI is both a search-model vendor and a hosted API provider, not a general chat catalog. Direct calls use a Jina Bearer key; this site exposes only adapted Embeddings and Rerank routes with a site-issued key.
- [Jina AI Embeddings, retrieval, and reranking](https://api.tu-zi.com/en/docs/providers/jina-ai/retrieval): Use this guide to understand the Jina AI Embeddings, retrieval, and reranking capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Kuaishou Kling AI](https://api.tu-zi.com/en/docs/providers/kling): Kuaishou Kling Open Platform provides image and video generation APIs for creative content production.
- [Kuaishou Kling AI Image generation and editing](https://api.tu-zi.com/en/docs/providers/kling/image): Use this guide to understand the Kuaishou Kling AI Image generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Kuaishou Kling AI Video generation and editing](https://api.tu-zi.com/en/docs/providers/kling/video): Use this guide to understand the Kuaishou Kling AI Video generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Meta Llama](https://api.tu-zi.com/en/docs/providers/meta-llama): Meta's open-weight model family, available for self-hosting or through partner cloud platforms.
- [Meta Llama Text, reasoning, and code](https://api.tu-zi.com/en/docs/providers/meta-llama/text): Use this guide to understand the Meta Llama Text, reasoning, and code capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [MiniMax](https://api.tu-zi.com/en/docs/providers/minimax): The mainland OpenAI-compatible base is `https://api.minimaxi.com/v1` and Anthropic compatibility uses `/anthropic`. Pay-as-you-go and Token Plan have separate keys, while this gateway uses its own token and adapted route.
- [MiniMax Speech and audio](https://api.tu-zi.com/en/docs/providers/minimax/audio): Use this guide to understand the MiniMax Speech and audio capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [MiniMax Image generation and editing](https://api.tu-zi.com/en/docs/providers/minimax/image): Use this guide to understand the MiniMax Image generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [MiniMax Text, reasoning, and code](https://api.tu-zi.com/en/docs/providers/minimax/text): Use this guide to understand the MiniMax Text, reasoning, and code capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [MiniMax Video generation and editing](https://api.tu-zi.com/en/docs/providers/minimax/video): Use this guide to understand the MiniMax Video generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Mistral AI](https://api.tu-zi.com/en/docs/providers/mistral): A first direct Mistral call uses the Mistral base URL, `Authorization: Bearer`, and an exact model ID. This site exposes only the adapted `/v1/chat/completions` route and requires this site's base URL and key. Batch, Guardrails, and Moderation are official Mistral capabilities, not evidence that this route automatically supports or enables them.
- [Mistral AI Speech and audio](https://api.tu-zi.com/en/docs/providers/mistral/audio): Use this guide to understand the Mistral AI Speech and audio capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Mistral AI Embeddings, retrieval, and reranking](https://api.tu-zi.com/en/docs/providers/mistral/retrieval): Use this guide to understand the Mistral AI Embeddings, retrieval, and reranking capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Mistral AI Text, reasoning, and code](https://api.tu-zi.com/en/docs/providers/mistral/text): Use this guide to understand the Mistral AI Text, reasoning, and code capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Moonshot Kimi](https://api.tu-zi.com/en/docs/providers/moonshot-kimi): Direct mainland China access uses `https://api.moonshot.cn/v1` and a Moonshot API key. K3, K2.6, and K2.7 use different thinking and tool parameters, while this gateway uses its own token and enabled model name.
- [Moonshot Kimi Text, reasoning, and code](https://api.tu-zi.com/en/docs/providers/moonshot-kimi/text): Use this guide to understand the Moonshot Kimi Text, reasoning, and code capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Ollama](https://api.tu-zi.com/en/docs/providers/ollama): Ollama provides both a local runtime and a cloud API, with different base URLs and authentication rules.
- [Ollama Embeddings, retrieval, and reranking](https://api.tu-zi.com/en/docs/providers/ollama/retrieval): Use this guide to understand the Ollama Embeddings, retrieval, and reranking capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Ollama Text, reasoning, and code](https://api.tu-zi.com/en/docs/providers/ollama/text): Use this guide to understand the Ollama Text, reasoning, and code capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [OpenAI](https://api.tu-zi.com/en/docs/providers/openai): OpenAI recommends Responses API for new projects. Direct and gateway calls both use a Bearer header, but require separate OpenAI and site keys, and the gateway endpoint must be implemented for the selected model.
- [OpenAI Speech and audio](https://api.tu-zi.com/en/docs/providers/openai/audio): Use this guide to understand the OpenAI Speech and audio capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [OpenAI Image generation and editing](https://api.tu-zi.com/en/docs/providers/openai/image): Use this guide to understand the OpenAI Image generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [OpenAI Embeddings, retrieval, and reranking](https://api.tu-zi.com/en/docs/providers/openai/retrieval): Use this guide to understand the OpenAI Embeddings, retrieval, and reranking capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [OpenAI Text, reasoning, and code](https://api.tu-zi.com/en/docs/providers/openai/text): Use this guide to understand the OpenAI Text, reasoning, and code capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [OpenAI Video generation and editing](https://api.tu-zi.com/en/docs/providers/openai/video): Use this guide to understand the OpenAI Video generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Alibaba Qwen](https://api.tu-zi.com/en/docs/providers/qwen): For direct Model Studio access, copy the API host for the target region and workspace and use a key from the same scope. New pay-as-you-go keys often start with sk-ws-, plan keys with sk-sp-, while this gateway uses its own token.
- [Alibaba Qwen Speech and audio](https://api.tu-zi.com/en/docs/providers/qwen/audio): Use this guide to understand the Alibaba Qwen Speech and audio capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Alibaba Qwen Image generation and editing](https://api.tu-zi.com/en/docs/providers/qwen/image): Use this guide to understand the Alibaba Qwen Image generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Alibaba Qwen Embeddings, retrieval, and reranking](https://api.tu-zi.com/en/docs/providers/qwen/retrieval): Use this guide to understand the Alibaba Qwen Embeddings, retrieval, and reranking capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Alibaba Qwen Text, reasoning, and code](https://api.tu-zi.com/en/docs/providers/qwen/text): Use this guide to understand the Alibaba Qwen Text, reasoning, and code capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Alibaba Qwen Video generation and editing](https://api.tu-zi.com/en/docs/providers/qwen/video): Use this guide to understand the Alibaba Qwen Video generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Tencent Hunyuan](https://api.tu-zi.com/en/docs/providers/tencent-hunyuan): For direct TokenHub access, call `/v1/models` and use the case-sensitive model ID. Custom online inference uses a service ID. Guangzhou and international bases differ, while this gateway uses its own token and adapted route.
- [Tencent Hunyuan Image generation and editing](https://api.tu-zi.com/en/docs/providers/tencent-hunyuan/image): Use this guide to understand the Tencent Hunyuan Image generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Tencent Hunyuan Embeddings, retrieval, and reranking](https://api.tu-zi.com/en/docs/providers/tencent-hunyuan/retrieval): Use this guide to understand the Tencent Hunyuan Embeddings, retrieval, and reranking capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Tencent Hunyuan Text, reasoning, and code](https://api.tu-zi.com/en/docs/providers/tencent-hunyuan/text): Use this guide to understand the Tencent Hunyuan Text, reasoning, and code capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Tencent Hunyuan Video generation and editing](https://api.tu-zi.com/en/docs/providers/tencent-hunyuan/video): Use this guide to understand the Tencent Hunyuan Video generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Vidu](https://api.tu-zi.com/en/docs/providers/vidu): Provides text-, image-, and reference-driven video generation alongside image and sound capabilities.
- [Vidu Image generation and editing](https://api.tu-zi.com/en/docs/providers/vidu/image): Use this guide to understand the Vidu Image generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Vidu Video generation and editing](https://api.tu-zi.com/en/docs/providers/vidu/video): Use this guide to understand the Vidu Video generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [xAI Grok](https://api.tu-zi.com/en/docs/providers/xai): A model platform for general generation, reasoning, coding, tool use, realtime voice, and media generation.
- [xAI Grok Image generation and editing](https://api.tu-zi.com/en/docs/providers/xai/image): Use this guide to understand the xAI Grok Image generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [xAI Grok Text, reasoning, and code](https://api.tu-zi.com/en/docs/providers/xai/text): Use this guide to understand the xAI Grok Text, reasoning, and code capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [xAI Grok Video generation and editing](https://api.tu-zi.com/en/docs/providers/xai/video): Use this guide to understand the xAI Grok Video generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [01.AI Yi](https://api.tu-zi.com/en/docs/providers/yi): Direct Yi access should use the current 01.AI console API key, base URL, and model ID. This gateway uses its own Bearer token and `/v1/chat/completions`; the credentials are not interchangeable.
- [01.AI Yi Text, reasoning, and code](https://api.tu-zi.com/en/docs/providers/yi/text): Use this guide to understand the 01.AI Yi Text, reasoning, and code capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Zhipu GLM](https://api.tu-zi.com/en/docs/providers/zhipu-glm): Direct GLM access uses `https://open.bigmodel.cn/api/paas/v4/`, a Bearer API key, and an exact model ID. This gateway uses its own token and guarantees only the adapted paths declared here.
- [Zhipu GLM Speech and audio](https://api.tu-zi.com/en/docs/providers/zhipu-glm/audio): Use this guide to understand the Zhipu GLM Speech and audio capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Zhipu GLM Image generation and editing](https://api.tu-zi.com/en/docs/providers/zhipu-glm/image): Use this guide to understand the Zhipu GLM Image generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Zhipu GLM Embeddings, retrieval, and reranking](https://api.tu-zi.com/en/docs/providers/zhipu-glm/retrieval): Use this guide to understand the Zhipu GLM Embeddings, retrieval, and reranking capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Zhipu GLM Text, reasoning, and code](https://api.tu-zi.com/en/docs/providers/zhipu-glm/text): Use this guide to understand the Zhipu GLM Text, reasoning, and code capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Zhipu GLM Video generation and editing](https://api.tu-zi.com/en/docs/providers/zhipu-glm/video): Use this guide to understand the Zhipu GLM Video generation and editing capability, then verify the exact callable model ID, endpoint, and price in the model marketplace before integration.
- [Tool use and AI agent model guide](https://api.tu-zi.com/en/docs/use-cases/agents): An agent is a controlled loop, not merely tool calling: structured schemas, least privilege, state, idempotency, timeouts, approval, audit, and replayable evaluation.
- [Unified async tasks: submit, poll, and retain results](https://api.tu-zi.com/en/docs/use-cases/async-tasks): For a suitable POST request, prefix its original path with `/async` and poll `/get-async?id=...`. Streaming, model discovery, Files, and Realtime cannot be wrapped this way.
- [Speech, transcription, and realtime audio guide](https://api.tu-zi.com/en/docs/use-cases/audio): Transcription, speech synthesis, and realtime voice are separate pipelines focused on recognition accuracy, voice rights, and low-latency duplex transport.
- [AI model API platform beginner guide: from signup to calls and billing checks](https://api.tu-zi.com/en/docs/use-cases/beginner-guide): Follow signup, marketplace verification, funding, a bounded API key, a minimal test, and Request-ID log lookup in that order to isolate problems and control spend.
- [Billing, usage, and cost reconciliation](https://api.tu-zi.com/en/docs/use-cases/billing-usage): Use the marketplace and current group for price. Retain model, path, Request-ID, usage or task ID, then reconcile the final consumption log instead of inferring cost from an early balance change.
- [Model originators, hosted clouds, aggregators, and local runtimes](https://api.tu-zi.com/en/docs/use-cases/channel-taxonomy): A provider guide answers who develops the model; a channel answers where a request is routed. The same model through another host can change authentication, protocol, region, version, and data boundary.
- [Choosing and calling coding models](https://api.tu-zi.com/en/docs/use-cases/coding): Select coding models by patch correctness, test pass rate, tool success, latency, and cost on real repositories, not marketplace copy or a single benchmark.
- [Doubao billing: online, Fast, batch, and cache](https://api.tu-zi.com/en/docs/use-cases/doubao-billing): Online inference serves interactive work, Fast is a model- and channel-dependent low-latency tier, and batch is a separate upstream service. Concurrent online requests do not receive batch discounts. Reconcile length, audio, cache, and the applicable group as well.
- [API endpoint and capability matrix](https://api.tu-zi.com/en/docs/use-cases/endpoint-matrix): Confirm the exact model and enabled endpoint in the marketplace first, then select a protocol here. A registered route only means the gateway recognizes it, not that every model or channel supports it.
- [Model evaluation, canarying, and version migration](https://api.tu-zi.com/en/docs/use-cases/evaluation-migration): Do not migrate from a leaderboard or one prompt. Pin the current model, parameters, and evaluation set, compare candidates, and canary quality, latency, errors, and cost before switching.
- [File and media input, output, and lifecycle](https://api.tu-zi.com/en/docs/use-cases/files-media): The gateway `/v1/files` routes are currently not implemented. Attach files in a form supported by the target model and endpoint, and treat every input or output URL as a permissioned resource with a lifecycle.
- [AI gateway admin runbook: channel rollout, retries, and supplier health](https://api.tu-zi.com/en/docs/use-cases/gateway-operations): Keep a new channel isolated while validating type, base URL, secret, models, and mapping; run a channel-specific test before canary traffic. Bound retries by count, time, replay safety, and health evidence.
- [AI API quickstart: from API key to first request](https://api.tu-zi.com/en/docs/use-cases/getting-started): An AI API call needs a matching base URL, API key, model ID, and endpoint; start with minimal curl, then add error handling, cost controls, and key security.
- [Image generation and vision model guide](https://api.tu-zi.com/en/docs/use-cases/image-generation): Vision understanding and image generation use different interfaces; generation projects must validate prompt adherence, editing consistency, text, likeness rights, moderation, and output terms.
- [Media and object-storage admin runbook: profiles, access, and lifecycle](https://api.tu-zi.com/en/docs/use-cases/media-storage-operations): Bind each purpose to an appropriate profile, test before saving, and sync lifecycle rules. Generated and staging data can expire by policy, while compliance and message assets need private delivery and independently approved retention.
- [AI API production reliability: timeouts, retries, limits, and SLOs](https://api.tu-zi.com/en/docs/use-cases/production-reliability): Reliable production calls need stage deadlines, jittered bounded retries, joint rate and concurrency protection, idempotency boundaries, tested fallback, and terminal-event SLOs.
- [RAG, embedding, and reranking model guide](https://api.tu-zi.com/en/docs/use-cases/rag): RAG combines chunking, embeddings, retrieval, reranking, grounded generation, and evaluation; completeness comes from citable evidence and no-answer handling, not a larger model list.
- [AI API security, privacy, and data governance](https://api.tu-zi.com/en/docs/use-cases/security-governance): Treat the model as an untrusted content processor: keep keys server-side, classify input, restrict URL fetching, reauthorize tool arguments, redact logs, and verify the data policy of the actual upstream.
- [SSE streaming and WebSocket realtime connections](https://api.tu-zi.com/en/docs/use-cases/streaming): SSE is an event stream in one HTTP response; WebSocket is a bidirectional session. A client must use protocol-specific terminal events instead of treating HTTP 200 or a closed connection as success.
- [General text, chat, and reasoning model guide](https://api.tu-zi.com/en/docs/use-cases/text-reasoning): General text models cover Q&A, writing, extraction, and analysis, while reasoning models fit multi-step problems; production selection should balance task accuracy, latency, cost, and control.
- [Common AI API error codes and troubleshooting](https://api.tu-zi.com/en/docs/use-cases/troubleshooting): Handle AI API failures by parsing the HTTP status, provider code, error type, message, retry hints, and request ID together. Fix request, authentication, billing, and hard-quota causes first; retry only transient throttling, network failures, and selected 5xx errors with strict bounds.
- [Video generation and asynchronous API guide](https://api.tu-zi.com/en/docs/use-cases/video-generation): Design video generation as an asynchronous job system with idempotent submit, persisted IDs, bounded polling or verified callbacks, terminal handling, output transfer, and moderation.