Finally, a FREE & No Authentication Perplexity AI API

An OpenAI-compatible API server that proxies requests through Perplexity AI. Any application or library that supports the OpenAI API format can connect to this server and use Perplexity's models without modification.

Built in Rust using Axum (just because) and curl-impersonate for TLS fingerprinting.

Features

• OpenAI API compatibility -- works with any OpenAI SDK or client out of the box
• Real-time streaming -- true server-sent events, not buffered chunking
• Image uploads -- send images via URL or base64 data URI in the image_url content type
• Multiple models -- access to GPT-5.2, Claude Sonnet 4.5, Claude Opus 4.6, Gemini 3, Grok 4.1, Kimi K2.5, and Sonar
• API key authentication -- optional Bearer token auth to secure your instance
• No Perplexity account required -- works anonymously for basic queries; add a session cookie for Pro features

Requirements

• Linux (x86_64)
• Rust 1.70+

The curl-impersonate binary is included in the curl/ directory.

Setup

cp .env.example .env

Edit .env to configure:

# Required: secures your API serverAPI_KEY=sk-your-secret-key-here# Optional: Perplexity session cookie for authenticated access# Without this, requests are made anonymously (still functional)PERPLEXITY_COOKIE=""# Optional: server port (default: 3030)PORT=3030

Getting a Perplexity cookie (optional)

A session cookie is only needed if you want authenticated access (Pro models, higher rate limits, image uploads on a free account). To get one:

1. Open perplexity.ai in your browser and log in
2. Open DevTools (F12) and go to the Network tab
3. Send any query on Perplexity
4. Find a request to perplexity.ai and copy the Cookie header value
5. Paste the full string into PERPLEXITY_COOKIE in your .env

OR

use a extension and output as a raw header string.

Without a cookie, the server runs in anonymous mode. Basic text queries still work. Image uploads require at minimum a free account cookie.

Build and Run

cargo build --release
cargo run --release

The server starts on http://localhost:3030 (or whatever PORT is set to).

API Endpoints

Method	Path	Description
GET	/health	Health check
POST	/v1/chat/completions	Chat completions
GET	/v1/models	List available models
GET	/v1/models/{model}	Get model details

Available Models

Model ID	Name
experimental	Sonar
gemini30flash	Gemini 3 Flash
gemini30pro	Gemini 3 Pro
gpt52	GPT-5.2
claude45sonnet	Claude Sonnet 4.5
claude45sonnetthinking	Claude Sonnet 4.5 (Thinking)
claude46opus	Claude Opus 4.6
grok41nonreasoning	Grok 4.1
kimik25thinking	Kimi K2.5

Usage

Python (OpenAI SDK)

fromopenaiimportOpenAIclient=OpenAI(
base_url="http://localhost:3030/v1",
api_key="sk-your-secret-key-here"
)
response=client.chat.completions.create(
model="gpt52",
messages=[{"role": "user", "content": "What is quantum computing?"}]
)
print(response.choices[0].message.content)

Streaming

stream=client.chat.completions.create(
model="gpt52",
messages=[{"role": "user", "content": "Explain general relativity"}],
stream=True
)
forchunkinstream:
content=chunk.choices[0].delta.contentifcontent:
print(content, end="", flush=True)

Image Input

response=client.chat.completions.create(
model="gpt52",
messages=[{
"role": "user",
"content": [
{"type": "text", "text": "Describe this image"},
{"type": "image_url", "image_url": {"url": "https://example.com/photo.jpg"}}
]
}]
)

Base64 data URIs are also supported in the url field.

curl

curl http://localhost:3030/v1/chat/completions \
-H "Authorization: Bearer sk-your-secret-key-here" \
-H "Content-Type: application/json" \
-d '{ "model": "gpt52", "messages": [{"role": "user", "content": "Hello"}] }'

Multi-turn Conversations

The API is stateless, following the OpenAI convention. The client maintains conversation history by including all previous messages in each request:

messages= [
{"role": "system", "content": "You are a helpful assistant."},
{"role": "user", "content": "What is Rust?"}
]
r1=client.chat.completions.create(model="gpt52", messages=messages)
messages.append({"role": "assistant", "content": r1.choices[0].message.content})
messages.append({"role": "user", "content": "Show me a hello world example"})
r2=client.chat.completions.create(model="gpt52", messages=messages)

Project Structure

src/
main.rs -- server setup, routing, auth middleware
openai_compat.rs -- OpenAI API translation layer
perplexity.rs -- Perplexity client, curl-impersonate integration
routes.rs -- health check endpoint
curl/
curl-impersonate -- patched curl binary with browser TLS fingerprinting
curl_chrome116 -- Chrome 116 wrapper script
.env.example -- configuration template
Cargo.toml -- dependencies

License

MIT