LiteLLM Perplexity Provider

Name: LiteLLM Perplexity Provider
Author: BerriAI

BerriAI April 23, 2026

0 copies 0 downloads

LiteLLM integration for Perplexity API providing a unified interface to call Perplexity Sonar models alongside 100+ other LLM providers with consistent API format, fallback routing, and cost tracking.

Agent Definition

## Overview

LiteLLM provides a unified Python/TypeScript interface to call Perplexity alongside 100+ other LLM providers. Use the same code to call Perplexity, OpenAI, Anthropic, and more with automatic fallback and load balancing.

## Supported Models

- perplexity/sonar
- perplexity/sonar-pro
- perplexity/sonar-reasoning-pro
- perplexity/sonar-deep-research

## Key Features

- Unified API across all providers
- Automatic fallback chains (e.g., Perplexity -> OpenAI -> Anthropic)
- Load balancing across models
- Cost tracking and budgeting
- Token usage monitoring
- OpenAI-compatible API format
- Streaming support

## Usage

```python
from litellm import completion

response = completion(
    model="perplexity/sonar-pro",
    messages=[{"role": "user", "content": "What's new in AI today?"}]
)
print(response.choices[0].message.content)
```

## Configuration

Set PERPLEXITY_API_KEY environment variable. LiteLLM automatically routes requests to Perplexity's API endpoint.

## Proxy Server

LiteLLM can run as a proxy server, exposing a unified OpenAI-compatible endpoint that routes to Perplexity and other providers based on model name.