首页 / LLM API / 配置指南
配置指南

按工具分别配置客户端

选择左侧列出的工具查看完整指南。下方内容直接使用您当前的API key和系统真实Base URL。

返回价格表 控制台

配置前准备

  • 在AI API控制台中充值并确保已创建API key。
  • 为每个客户端使用正确的Base URL:有些需要`/v1`,有些不需要。
  • 在价格表中选择当前活跃的模型,避免已关闭路由的模型。
  • 如果客户端支持自定义endpoint,始终使用本站点的HTTPS endpoint。

Base URL规则

  • OpenAI兼容客户端:使用带`/v1`的Base URL。
  • Claude Code环境变量`ANTHROPIC_BASE_URL`:使用根域名,不要添加`/v1`。
API reference

Endpoints and commands

Open an endpoint to see its complete curl command. The list stays compact instead of rendering every command at once.

Text / hội thoại

chat/completionsPOST

OpenAI-compatible chat; messages, tools and optional stream=true.

https://api.xah.io/v1/chat/completions

Complete command
API_KEY="YOUR_CKEY_API_KEY"

curl https://api.xah.io/v1/chat/completions \
  -H "Authorization: Bearer $API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{
  "model": "MODEL",
  "messages": [{"role": "user", "content": "Hello"}],
  "stream": false
}'
completionsPOST

Legacy prompt completion; use only when this capability is enabled.

https://api.xah.io/v1/completions

Complete command
API_KEY="YOUR_CKEY_API_KEY"

curl https://api.xah.io/v1/completions \
  -H "Authorization: Bearer $API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{
  "model": "MODEL",
  "prompt": "Hello",
  "max_tokens": 256
}'
responsesPOST

Responses-style request; input/output follows the enabled upstream contract.

https://api.xah.io/v1/responses

Complete command
API_KEY="YOUR_CKEY_API_KEY"

curl https://api.xah.io/v1/responses \
  -H "Authorization: Bearer $API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{
  "model": "MODEL",
  "input": "Hello"
}'
messagesPOST

Anthropic Messages format. Send anthropic-version: 2023-06-01 and max_tokens.

https://api.xah.io/v1/messages

Complete command
API_KEY="YOUR_CKEY_API_KEY"

curl https://api.xah.io/v1/messages \
  -H "Authorization: Bearer $API_KEY" \
  -H 'Content-Type: application/json' \
  -H 'anthropic-version: 2023-06-01' \
  -d '{
  "model": "MODEL",
  "max_tokens": 1024,
  "messages": [{"role": "user", "content": "Hello"}]
}'

Images

images/generationsPOST

Text-to-image. Use only fields documented by the selected image model: prompt, n, size, quality, response_format, etc.

https://api.xah.io/v1/images/generations

Complete command
API_KEY="YOUR_CKEY_API_KEY"

curl https://api.xah.io/v1/images/generations \
  -H "Authorization: Bearer $API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{
  "model": "MODEL",
  "prompt": "A red fox in snow",
  "n": 1,
  "size": "1024x1024"
}'
images/generations · image-to-imagePOST

Tạo ảnh từ ảnh cũng gọi cùng endpoint images/generations. Gửi ảnh theo field mà model/upstream đã công bố trong payload template; ví dụ dưới dùng image_url là contract của model mẫu.

https://api.xah.io/v1/images/generations

Complete command
API_KEY="YOUR_CKEY_API_KEY"

curl https://api.xah.io/v1/images/generations \
  -H "Authorization: Bearer $API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{
  "model": "MODEL",
  "prompt": "Turn the reference into a watercolor illustration",
  "image_url": "https://example.com/reference.png",
  "n": 1,
  "size": "1024x1024"
}'
images/editsPOST multipart

Image-to-image edit/generation. image is multipart; prompt and mask depend on the selected model contract.

https://api.xah.io/v1/images/edits

Complete command
API_KEY="YOUR_CKEY_API_KEY"

curl https://api.xah.io/v1/images/edits \
  -H 'Authorization: Bearer API_KEY' \
  -F 'model=MODEL' \
  -F 'prompt=Make it brighter' \
  -F '[email protected]'
images/variationsPOST multipart

Create variations from an input image. image is multipart; do not send JSON/base64 unless the selected upstream documents it.

https://api.xah.io/v1/images/variations

Complete command
API_KEY="YOUR_CKEY_API_KEY"

curl https://api.xah.io/v1/images/variations \
  -H 'Authorization: Bearer API_KEY' \
  -F 'model=MODEL' \
  -F '[email protected]'

Audio

audio/speechPOST

Text-to-speech using the official OpenAI Audio API contract. JSON body: model, input, voice, optional instructions, response_format, speed and stream_format.

https://api.xah.io/v1/audio/speech

Complete command
API_KEY="YOUR_CKEY_API_KEY"

curl https://api.xah.io/v1/audio/speech \
  -H "Authorization: Bearer $API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{
  "model": "MODEL",
  "input": "Hello from CKEY",
  "voice": "alloy",
  "response_format": "mp3",
  "speed": 1.0
}'
audio/transcriptionsPOST multipart

Speech-to-text using the official OpenAI contract. Multipart fields: file, model, optional language (ISO-639-1), prompt and response_format.

https://api.xah.io/v1/audio/transcriptions

Complete command
API_KEY="YOUR_CKEY_API_KEY"

curl https://api.xah.io/v1/audio/transcriptions \
  -H 'Authorization: Bearer API_KEY' \
  -F '[email protected]' -F 'model=gpt-4o-transcribe' -F 'language=en' -F 'response_format=json'
audio/translationsPOST multipart

Official OpenAI translation contract. Only whisper-1 is supported; endpoint and model are deprecated.

https://api.xah.io/v1/audio/translations

Complete command
API_KEY="YOUR_CKEY_API_KEY"

curl https://api.xah.io/v1/audio/translations \
  -H 'Authorization: Bearer API_KEY' \
  -F '[email protected]' -F 'model=whisper-1' -F 'response_format=json'

Video (task / async)

v1/videosPOST + GET

OpenAI-compatible Video API V1. POST creates a job at /v1/videos/generations; poll GET /v1/videos/{JOB_ID}. Use seconds for the V1 contract.

https://api.xah.io/v1/videos/generations

Complete command
API_KEY="YOUR_CKEY_API_KEY"

curl https://api.xah.io/v1/videos/generations \
  -H "Authorization: Bearer $API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{
  "model": "MODEL",
  "prompt": "A gentle camera move across a mountain",
  "seconds": 8
}

# Poll: GET https://api.xah.io/v1/videos/JOB_ID'
v2/video_generationPOST + GET + DELETE

CKEY public Task API V2. POST creates a task; poll GET /v2/query/video_generation/{JOB_ID}; cancel DELETE /v2/video_generation/{JOB_ID}. Use duration for the V2 contract; do not change the route to V1.

https://api.xah.io/v2/video_generation

Complete command
API_KEY="YOUR_CKEY_API_KEY"

curl https://api.xah.io/v2/video_generation \
  -H "Authorization: Bearer $API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{
  "model": "MODEL",
  "content": [{"type": "text", "text": "A gentle camera move across a mountain"}],
  "duration": 6,
  "ratio": "16:9"
}

# Poll: GET https://api.xah.io/v2/query/video_generation/JOB_ID
# Cancel: DELETE https://api.xah.io/v2/video_generation/JOB_ID'

Embeddings / ranking / safety / search

embeddingsPOST

Convert one string or an array of strings to vectors.

https://api.xah.io/v1/embeddings

Complete command
API_KEY="YOUR_CKEY_API_KEY"

curl https://api.xah.io/v1/embeddings \
  -H "Authorization: Bearer $API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{
  "model": "MODEL",
  "input": ["first text", "second text"]
}'
rerankPOST

Rank documents for a query. query and documents are required; top_n must not exceed document count.

https://api.xah.io/v1/rerank

Complete command
API_KEY="YOUR_CKEY_API_KEY"

curl https://api.xah.io/v1/rerank \
  -H "Authorization: Bearer $API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{
  "model": "MODEL",
  "query": "latest AI news",
  "documents": ["doc A", "doc B"],
  "top_n": 2
}'
moderationsPOST

Classify input using the enabled moderation model.

https://api.xah.io/v1/moderations

Complete command
API_KEY="YOUR_CKEY_API_KEY"

curl https://api.xah.io/v1/moderations \
  -H "Authorization: Bearer $API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{
  "model": "MODEL",
  "input": "Text to classify"
}'
searchPOST

Provider/model-specific search contract. Send optional fields only when the selected model documents them.

https://api.xah.io/v1/search

Complete command
API_KEY="YOUR_CKEY_API_KEY"

curl https://api.xah.io/v1/search \
  -H "Authorization: Bearer $API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{
  "model": "MODEL",
  "query": "latest AI news",
  "max_results": 5
}'

Native provider routes

geminiPOST

Google Gemini native format. Streaming uses :streamGenerateContent?alt=sse, not an invented stream body field.

https://api.xah.io/v1beta/models/MODEL:generateContent

Complete command
API_KEY="YOUR_CKEY_API_KEY"

curl https://api.xah.io/v1beta/models/MODEL:generateContent \
  -H "Authorization: Bearer $API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{
  "contents": [{"role": "user", "parts": [{"text": "Hello"}]}]
}'
ollamaPOST

Ollama native chat format.

https://api.xah.io/api/chat

Complete command
API_KEY="YOUR_CKEY_API_KEY"

curl https://api.xah.io/api/chat \
  -H "Authorization: Bearer $API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{
  "model": "MODEL",
  "messages": [{"role": "user", "content": "Hello"}],
  "stream": false
}'
api/generatePOST

Ollama native prompt generation.

https://api.xah.io/api/generate

Complete command
API_KEY="YOUR_CKEY_API_KEY"

curl https://api.xah.io/api/generate \
  -H "Authorization: Bearer $API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{
  "model": "MODEL",
  "prompt": "Hello",
  "stream": false
}'
Contract rule

Both text-to-image and image-to-image use POST /images/generations when that is the model contract. Do not rename the image field: use the payload template/contract of the selected model. Do not substitute images/edits or images/variations for a model that exposes only images/generations.

OpenAI-compatible chat/completions /v1

Cursor

Cursor应使用自定义API key和自定义base URL走OpenAI兼容模式。

需要填写的信息

Provider
OpenAI-compatible
API Key
sk-YOUR_API_KEY
Base URL
https://api.xah.io/v1
模型
gpt-5.5
路由
chat/completions
套餐要求
需要支持自定义API endpoint的Cursor套餐。

建议步骤

  1. 1打开Cursor Settings,进入AI / Models / API Keys。
  2. 2如果您的Cursor版本使用其他名称,请启用自定义OpenAI key或自定义provider选项。
  3. 3粘贴下方显示的API key和Base URL。
  4. 4选择推荐模型或价格表中其他当前活跃的模型。

可直接复制的配置模板

Provider: OpenAI-compatible
API Key: sk-YOUR_API_KEY
Base URL: https://api.xah.io/v1
Model: gpt-5.5
Route: chat/completions

推荐尝试的模型

  • gpt-5.5
  • claude-opus-4.6
  • claude-opus-4.7
  • claude-opus-4.8
  • claude-sonnet-4.6
  • deepseek-v4.1-flash

注意事项

  • Cursor使用OpenAI兼容模式,因此Base URL必须包含`/v1`。
  • 如果某个模型在Cursor中无法运行,请切换到列表中其他活跃的OpenAI兼容模型。
  • 保存后,发送一个简短prompt确认`chat/completions`路由正常工作。
联系支持
我们随时为您提供帮助