One endpoint. OpenAI & Anthropic compatible.

Quick Start

Point any OpenAI- or Anthropic-compatible client at this server and authenticate with your API key (the master key from .env, or any key created on the dashboard).

curl https://deepsegs.irengcloud.com/v1/chat/completions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "expert",
    "messages": [{"role": "user", "content": "Hello!"}],
    "stream": false
  }'

curl https://deepsegs.irengcloud.com/v1/messages \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "expert",
    "max_tokens": 1024,
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

Set "stream": true for server-sent events. Also available: POST /v1/responses, GET /v1/models, POST /v1/files, GET /v1/files/{id}/content, POST /v1/files/upload.

With the OpenAI SDK

from openai import OpenAI

client = OpenAI(base_url="https://deepsegs.irengcloud.com/v1", api_key="YOUR_API_KEY")

resp = client.chat.completions.create(
    model="expert",
    messages=[{"role": "user", "content": "Hello!"}],
)
print(resp.choices[0].message.content)

Models

Aliases anthropic/claude-instant, anthropic/claude-vision, anthropic/claude-expert are also exposed for Claude Desktop auto-discovery. If no model is sent, requests default to expert.

Setup: DeepSeek Auth Token

1. Open an incognito/private window and log in at chat.deepseek.com.
2. Open DevTools (F12) → Console, then run: JSON.parse(localStorage.getItem("userToken")).value
3. Copy the raw token (no quotes) into the dashboard.
4. Close the incognito window to keep the session alive. Logging out of DeepSeek in a browser invalidates the token.

Features