Perplexity OpenAI-Compatible API Server

Transform Perplexity AI into a drop-in replacement for OpenAI's API. This server bridges the gap between Perplexity's powerful search-augmented intelligence and applications built for OpenAI's standard interface. Deploy in seconds with Docker, no code changes required. It is forked from henrique-coder/perplexity-webui-scraper and uses Python and FastAPI to create a RESTful API server.

Features

• Models are automatically discovered from Perplexity.
• One-click deployment with Docker
• Request rate limiting

Quick Start

Docker (Recommended)

# 1. Copy environment template
cp .env.example .env

# 2. Add your Perplexity session token to .env

# 3. Start the server
docker-compose up -d

# 4. Test
curl http://localhost:8000/health
curl http://localhost:8000/v1/models

Manual

# Install dependencies
pip install -r requirements.txt
pip install -e .

# Copy and configure .env
cp .env.example .env
# Edit .env with your session token

# Run
python openai_server.py

Getting Your Session Token

1. Log in at perplexity.ai
2. Open DevTools (F12) → Application → Cookies
3. Copy __Secure-next-auth.session-token value
4. Add to .env: PERPLEXITY_SESSION_TOKEN=your_token

API Usage

The server is 100% OpenAI API compatible:

import openai

client = openai.OpenAI(
    api_key="your-api-key",  # Optional
    base_url="http://localhost:8000/v1"
)

response = client.chat.completions.create(
    model="perplexity-auto",
    messages=[{"role": "user", "content": "Hello!"}]
)

print(response.choices[0].message.content)

curl -X POST http://localhost:8000/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{"model": "perplexity-auto", "messages": [{"role": "user", "content": "Hello!"}]}'

Endpoints

Endpoint	Method	Description
/v1/chat/completions	POST	Chat completions
/v1/models	GET	List available models
/v1/models/refresh	POST	Refresh models from Perplexity
/conversations	GET	List conversations
/stats	GET	Server statistics
/health	GET	Health check

Configuration

Variable	Default	Description
PERPLEXITY_SESSION_TOKEN	-	Required: Session token
OPENAI_API_KEY	-	Optional: API key for auth
PORT	8000	Server port
LOG_LEVEL	INFO	Logging level
ENABLE_RATE_LIMITING	true	Enable rate limiting
REQUESTS_PER_MINUTE	60	Rate limit
CONVERSATION_TIMEOUT	3600	Session timeout (seconds)
DEFAULT_MODEL	perplexity-auto	Default model

Models

Models are automatically fetched from Perplexity. Common models include:

• perplexity-auto - Auto-select best model
• perplexity-sonar - Fast responses
• perplexity-research - Deep research
• GPT, Claude, Gemini, Grok models via Perplexity

Use /v1/models to see all available models.

License

MIT

Disclaimer

This is an unofficial implementation using internal Perplexity APIs. Use at your own risk.