Multi-Provider API Gateway

Exocore AI Gateway

Unified REST API for Gemini, Claude, DeepSeek, Qwen, and Meta AI — all behind a single server powered by @exocore/multi.

Base URL

Google Gemini

Gemini API

Cookie-based proxy to Google Gemini — multi-turn chat, file uploads, and session management.

Status

—

last checked

—

validity

Client

—

initialized

cURL

curl "/api/gemini/health"

cURL

curl "/api/gemini/cookies"

Replace the cookie jar with a fresh raw cookie string. The server parses, saves, and re-initializes the Gemini client.

Field	Type	Required	Description
body	string	Required	Raw cookie string `SID=x; HSID=y; ...`

cURL

curl -X POST "/api/gemini/cookies" \
  -H "Content-Type: text/plain" \
  --data-raw "SID=abc; HSID=xyz; ..."

Cookie string

cURL

curl -X DELETE "/api/gemini/cookies"

Sends a text prompt to Gemini. Supports multi-turn with cid, rid, rcid params.

Param	Type	Req	Description
ask	string	Required*	Text prompt
upload	string	Opt	File URLs to attach
cid	string	Opt	Conversation ID
rid	string	Opt	Response ID
rcid	string	Opt	Candidate ID
model	string	Opt	`fast`, `thinking`, or `pro` (default: `fast`)
tokens	bool	Opt	Stream tokens as SSE (`true`)
delete	bool	Opt	Auto-delete conversation after reply

cURL

curl "/api/gemini/chat?ask=What+is+the+capital+of+France"

Message

Model

Delete after chat

Stream tokens

Multi-turn IDs (optional)

cid

rid

rcid

Truly temporary Gemini chat — no cookies, no auth required. Sessions are ephemeral.

cURL

curl "/api/gemini/temp/chat?ask=Hello"

Message

Conversations

cURL

curl "/api/gemini/conversations"

cURL

curl -X DELETE "/api/gemini/conversations"

cURL

curl -X DELETE "/api/gemini/conversations/<id>"

Conversation ID

DeepSeek

DeepSeek API

Token-based client for DeepSeek chat. Configure via deepseek.json with userToken and hifLeim fields.

Status

—

configured

cURL

curl "/api/deepseek/health"

Sends a message to DeepSeek via GET query. Auto-creates session if sessionId omitted.

Param	Type	Req	Description
prompt	string	Req	Your message
sessionId	string	Opt	Resume existing session
thinkingEnabled	bool	Opt	Enable reasoning
searchEnabled	bool	Opt	Enable web search
delete	bool	Opt	Auto-delete session after reply
tokens	bool	Opt	Stream tokens as SSE (`true`)

cURL

curl "/api/deepseek/chat?prompt=Hello"

Message

Delete after chat

Stream tokens

Same as GET but with JSON body.

cURL

curl -X POST "/api/deepseek/chat" \
  -H "Content-Type: application/json" \
  --data '{"prompt":"Hello"}'

Message

Stream tokens

Sessions

cURL

curl "/api/deepseek/sessions?limit=20"

cURL

curl -X POST "/api/deepseek/sessions"

cURL

curl -X DELETE "/api/deepseek/sessions"

cURL

curl -X DELETE "/api/deepseek/sessions/<id>"

Session ID

cURL

curl "/api/deepseek/sessions/<id>/messages"

Session ID

Send a message to an existing session.

cURL

curl -X POST "/api/deepseek/sessions/<id>/chat" \
  -H "Content-Type: application/json" \
  --data '{"prompt":"Hello again"}'

Session ID

Message

Regenerate an assistant response for a given message.

cURL

curl -X POST "/api/deepseek/sessions/<id>/regenerate/<msgId>"

Session ID

Message ID

Edit a user message and re-generate the response.

cURL

curl -X POST "/api/deepseek/sessions/<id>/edit/<msgId>" \
  -H "Content-Type: application/json" \
  --data '{"prompt":"edited text"}'

Session ID

Message ID

New prompt

Continue generating from a partial assistant response.

cURL

curl -X POST "/api/deepseek/sessions/<id>/continue/<msgId>"

Session ID

Message ID

Stop an in-progress streaming response.

cURL

curl -X POST "/api/deepseek/sessions/<id>/stop/<msgId>"

Session ID

Message ID

cURL

curl -X PUT "/api/deepseek/sessions/<id>/title" \
  -H "Content-Type: application/json" \
  --data '{"title":"My Chat"}'

Session ID

Title

cURL

curl -X PUT "/api/deepseek/sessions/<id>/pinned" \
  -H "Content-Type: application/json" \
  --data '{"pinned":true}'

Session ID

Pinned

User & PoW

cURL

curl "/api/deepseek/user"

Request a PoW challenge for a target endpoint.

Param	Type	Req	Description
target	string	Opt	Default `/api/v0/chat/completion`

cURL

curl "/api/deepseek/pow-challenge"

Solve a PoW challenge and get the answer.

cURL

curl -X POST "/api/deepseek/pow-solve" \
  -H "Content-Type: application/json" \
  --data '{"challenge":"...","difficulty":...}'

cURL

curl "/api/deepseek/shares"

cURL

curl -X POST "/api/deepseek/shares" \
  -H "Content-Type: application/json" \
  --data '{"sessionId":"...","messageIds":["..."]}'

Session ID

Message IDs (comma)

Fork a shared conversation into a new session.

cURL

curl -X POST "/api/deepseek/shares/fork" \
  -H "Content-Type: application/json" \
  --data '{"shareId":"..."}'

Share ID

cURL

curl -X DELETE "/api/deepseek/shares/<id>"

Share ID

Qwen

Qwen API

Token-based client for Qwen AI chat. Configure via qwen.json cookies file with a token entry.

Status

—

configured

cURL

curl "/api/qwen"

cURL

curl "/api/qwen/health"

Send a prompt to Qwen via GET. Supports modes: normal, thinking, fast. Use chat_type for features: slides, travel, learn, deep_research, web_dev, search, artifacts.

Param	Type	Req	Description
prompt	string	Req	Your message
mode	string	Opt	`normal`, `thinking`, or `fast`
image_url	string	Opt	URL for vision queries
chat_type	string	Opt	Feature: slides, travel, deep_research, etc.
model	string	Opt	Override model name
delete	bool	Opt	Auto-delete all conversations after reply
tokens	bool	Opt	Stream tokens as SSE (`true`)

cURL

curl "/api/qwen/ask?prompt=Hello&mode=fast"

Message

Mode

Delete after chat

Stream tokens

POST version with JSON body. Supports file attachments via file_path.

cURL

curl -X POST "/api/qwen/ask" \
  -H "Content-Type: application/json" \
  --data '{"prompt":"Hello","mode":"fast","chat_type":"slides"}'

Message

Mode

Stream tokens

Features

cURL

curl "/api/qwen/ask?prompt=3+slides+about+AI&chat_type=slides"

Message

cURL

curl "/api/qwen/ask?prompt=Tokyo+itinerary&chat_type=travel"

Message

cURL

curl "/api/qwen/ask?prompt=explain+quantum&chat_type=learn"

Message

cURL

curl "/api/qwen/ask?prompt=climate+change+effects&chat_type=deep_research"

Message

cURL

curl "/api/qwen/ask?prompt=React+component&chat_type=web_dev"

Message

cURL

curl "/api/qwen/ask?prompt=latest+tech+news&chat_type=search"

Message

cURL

curl "/api/qwen/ask?prompt=draw+an+SVG+logo&chat_type=artifacts"

Message

cURL

curl "/api/qwen/ask?prompt=calculator+tool&chat_type=tools"

Message

cURL

curl "/api/qwen/ask?prompt=Python+fibonacci&chat_type=code"

Message

cURL

curl "/api/qwen/ask?prompt=design+a+UI+sketch&chat_type=drawing"

Message

cURL

curl "/api/qwen/ask?prompt=you+are+a+detective&chat_type=roleplay"

Message

Files & Conversations

Upload a file as base64 content.

cURL

curl -X POST "/api/qwen/upload" \
  -H "Content-Type: application/json" \
  --data '{"content":"aGVsbG8=","filename":"hello.txt"}'

Retrieve an uploaded file by filename.

cURL

curl "/api/qwen/uploads/<filename>"

cURL

curl "/api/qwen/conversations"

cURL

curl -X DELETE "/api/qwen/conversations"

cURL

curl -X DELETE "/api/qwen/conversations/<id>"

Conversation ID

Kimi

Kimi API

Token-based client for Kimi AI chat. Configure via kimi.json cookies file with a kimi-auth cookie entry (JWT token).

Status

—

configured

cURL

curl "/api/kimi/health"

cURL

curl "/api/kimi/user"

cURL

curl "/api/kimi/models"

Chat

Params: prompt (req), sessionId, thinking, delete (Auto-delete session after reply).

cURL

curl "/api/kimi/chat?prompt=Hello"

Message

Delete after chat

Send message via POST with JSON body.

cURL

curl -X POST "/api/kimi/chat" \
  -H "Content-Type: application/json" \
  --data '{"prompt":"Hello"}'

Sessions

cURL

curl "/api/kimi/sessions"

cURL

curl -X POST "/api/kimi/sessions" \
  -H "Content-Type: application/json" \
  --data '{"name":"My Chat"}'

cURL

curl "/api/kimi/sessions/<id>"

cURL

curl "/api/kimi/sessions/<id>/messages"

Send message to an existing session.

cURL

curl -X POST "/api/kimi/chat/<id>" \
  -H "Content-Type: application/json" \
  --data '{"prompt":"Hello"}'

cURL

curl -X PUT "/api/kimi/sessions/<id>" \
  -H "Content-Type: application/json" \
  --data '{"chat":{"name":"New Name"}}'

cURL

curl -X DELETE "/api/kimi/sessions/<id>"

cURL

curl -X DELETE "/api/kimi/sessions"

Kimi+

cURL

curl "/api/kimi/kimiplus"

cURL

curl "/api/kimi/kimiplus/<id>"

Other

cURL

curl -X POST "/api/kimi/cancel/<chatId>/<messageId>"

cURL

curl -X POST "/api/kimi/vote/<chatId>/<messageId>" \
  -H "Content-Type: application/json" \
  --data '{"voteType":"upvote"}'

Anthropic Claude

Claude API

Free Claude access via Puter's driver API. Configure an auth_token in data/claude.json.

Status

—

configured

cURL

curl "/api/claude/health"

cURL

curl "/api/claude/models"

Chat

Query params: ask (required), model (optional, default: claude-sonnet-4-6).

cURL

curl "/api/claude/chat?ask=Hello"

Message

Send messages array with optional streaming. Body: { messages, model?, stream?, max_tokens? }.

cURL

curl -X POST "/api/claude/chat" \
  -H "Content-Type: application/json" \
  --data '{"messages":[{"role":"user","content":"Hello"}],"model":"claude-sonnet-4-6"}'

Available Models

claude-sonnet-4-6

claude-opus-4-7

claude-haiku-4-5

claude-opus-4-6

claude-3-sonnet

claude-3-haiku

claude-3-opus

V1 / Anthropic-compatible

Claude also supports the /v1/messages endpoint with Anthropic-style model names — the gateway maps them to Puter driver models automatically.

Anthropic Message API format. Model names like claude-sonnet-4-20250514, claude-opus-4-7 are mapped automatically.

cURL

curl -X POST "/v1/messages" \
  -H "Content-Type: application/json" \
  --data '{"model":"claude-sonnet-4-20250514","messages":[{"role":"user","content":"Hello"}],"max_tokens":1024}'

OpenAI

GPT API

Free GPT access via Puter's driver API. Configure an auth_token in data/gpt.json (can reuse data/claude.json token).

Status

—

configured

cURL

curl "/api/gpt/health"

cURL

curl "/api/gpt/models"

Chat

Query params: ask (required), model (optional, default: gpt-4o).

cURL

curl "/api/gpt/chat?ask=Hello"

Message

Send messages array with optional streaming. Body: { messages, model?, stream?, max_tokens? }.

cURL

curl -X POST "/api/gpt/chat" \
  -H "Content-Type: application/json" \
  --data '{"messages":[{"role":"user","content":"Hello"}],"model":"gpt-4o"}'

Available Models

gpt-4o

gpt-4o-mini

gpt-4.1

gpt-4.1-mini

gpt-4.1-nano

gpt-4.5-preview

gpt-5-chat-latest

gpt-5.1

gpt-5.1-chat-latest

gpt-5.2-chat-latest

gpt-5.4

gpt-5.4-nano

gpt-5.5

o1-mini

o3-mini

o4-mini

xAI

Grok API

Free Grok access via Puter's driver API. Configure an auth_token in data/grok.json (can reuse data/claude.json token).

Status

—

configured

cURL

curl "/api/grok/health"

cURL

curl "/api/grok/models"

Chat

Query params: ask (required), model (optional, default: grok-3).

cURL

curl "/api/grok/chat?ask=Hello"

Message

Send messages array with optional streaming. Body: { messages, model?, stream?, max_tokens? }.

cURL

curl -X POST "/api/grok/chat" \
  -H "Content-Type: application/json" \
  --data '{"messages":[{"role":"user","content":"Hello"}],"model":"grok-3"}'

Available Models

grok-3

grok-3-fast

grok-3-mini

grok-3-mini-fast

grok-4

grok-4-fast

grok-4-1-fast

grok-4.3

grok-beta

grok-2-vision

Meta AI API

Cookie-based proxy to Meta AI (Llama). Requires ecto_1_sess and datr cookies in data/meta.json.

Status

—

configured

cURL

curl "/api/meta/health"

cURL

curl "/api/meta/models"

Chat

Query params: ask (required). Optionally delete=true to auto-delete conversation.

cURL

curl "/api/meta/chat?ask=Hello"

Message

Delete after chat

Send messages array. Body: { messages, model? }.

cURL

curl -X POST "/api/meta/chat" \
  -H "Content-Type: application/json" \
  --data '{"messages":[{"role":"user","content":"Hello"}]}'

Image Generation

Query params: prompt (required), orientation (optional: VERTICAL, HORIZONTAL, SQUARE).

cURL

curl "/api/meta/image?prompt=a+cat&orientation=SQUARE"

Prompt

Orientation

Video Generation

Query params: prompt (required). Video generation is async — returns a processing status.

cURL

curl "/api/meta/video?prompt=a+cat+playing"

Prompt

Available Models

meta-ai

OpenAI-Compatible

V1 API

Unified OpenAI-compatible /v1/chat/completions endpoint that routes to any provider. All models are mapped automatically — just set the model field in your request body.

Endpoint

POST /v1/chat/completions

OpenAI-compatible

Available Models

gemini

gemini-fast

gemini-flash

gemini-thinking

gemini-pro

deepseek

deepseek-default

deepseek-instant

deepseek-thinking

deepseek-expert

deepseek-search

deepseek-vision

qwen

qwen-normal

qwen-fast

qwen-thinking

qwen3.7-max

qwen3.7-max-fast

qwen3.7-max-thinking

qwen3.6-plus

qwen3.6-plus-fast

qwen3.6-plus-thinking

qwen3.5-flash

qwen3.5-flash-normal

qwen3.5-flash-thinking

Usage Examples

Gemini

cURL

curl -X POST "/v1/chat/completions" \
  -H "Content-Type: application/json" \
  --data '{"model":"gemini-thinking","messages":[{"role":"user","content":"What is 2+2?"}]}'

DeepSeek

cURL

curl -X POST "/v1/chat/completions" \
  -H "Content-Type: application/json" \
  --data '{"model":"deepseek-thinking","messages":[{"role":"user","content":"Explain quantum computing"}]}'

Qwen

cURL

curl -X POST "/v1/chat/completions" \
  -H "Content-Type: application/json" \
  --data '{"model":"qwen-thinking","messages":[{"role":"user","content":"Write a poem"}]}'

Kimi

cURL

curl -X POST "/v1/chat/completions" \
  -H "Content-Type: application/json" \
  --data '{"model":"kimi-thinking","messages":[{"role":"user","content":"Hello"}]}'

Claude

cURL

curl -X POST "/v1/chat/completions" \
  -H "Content-Type: application/json" \
  --data '{"model":"claude-opus-4-7","messages":[{"role":"user","content":"Explain relativity"}]}'

Meta AI

cURL

curl -X POST "/v1/chat/completions" \
  -H "Content-Type: application/json" \
  --data '{"model":"meta","messages":[{"role":"user","content":"Hello"}]}'

GPT (OpenAI)

cURL

curl -X POST "/v1/chat/completions" \
  -H "Content-Type: application/json" \
  --data '{"model":"gpt-4o","messages":[{"role":"user","content":"Hello"}]}'

Grok (xAI)

cURL

curl -X POST "/v1/chat/completions" \
  -H "Content-Type: application/json" \
  --data '{"model":"grok-4","messages":[{"role":"user","content":"Hello"}]}'

Anthropic-compatible

The gateway also supports Anthropic's /v1/messages format with model override via URL path prefix.

Anthropic-style message format. Maps model names like claude-sonnet-4-20250514, claude-opus-4-7 etc. automatically.

cURL

curl -X POST "/v1/messages" \
  -H "Content-Type: application/json" \
  --data '{"model":"claude-sonnet-4-20250514","messages":[{"role":"user","content":"Hello"}],"max_tokens":1024}'

Override the model by prefixing the URL path. Works with any V1 model ID.

cURL

curl -X POST "/claude-sonnet/v1/messages" \
  -H "Content-Type: application/json" \
  --data '{"messages":[{"role":"user","content":"Hello"}],"max_tokens":1024}'

Exocore AI Gateway

Gemini API

Conversations

DeepSeek API

Sessions

User & PoW

Shares

Qwen API

Features

Files & Conversations

Kimi API

Chat

Sessions

Kimi+

Other

Claude API

Chat

Available Models

V1 / Anthropic-compatible

GPT API

Chat

Available Models

Grok API

Chat

Available Models

Meta AI API

Chat

Image Generation

Video Generation

Available Models

V1 API

Available Models

Usage Examples

Gemini

DeepSeek

Qwen

Kimi

Claude

Meta AI

GPT (OpenAI)

Grok (xAI)

Anthropic-compatible