Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.perplexity.ai/llms.txt

Use this file to discover all available pages before exploring further.

​
Overview

LiteLLM is a Python SDK and proxy server that gives you a single OpenAI-compatible interface to 100+ LLM providers. Both Perplexity’s Sonar models and the Agent API (with third-party models like GPT-5, Claude, and Gemini routed through Perplexity) are first-class providers in LiteLLM.
LiteLLM lets you swap providers without rewriting code, run a self-hosted proxy that fronts every model behind one API key, and track spend, latency, and errors per provider. Learn more at litellm.ai.

​
Installation

pip install litellm

​
API Key Setup

LiteLLM uses two environment variables depending on which Perplexity endpoint you’re calling: 
# For Sonar chat completions (litellm.completion)
export PERPLEXITYAI_API_KEY="your_api_key_here"

# For Agent API responses (litellm.responses)
export PERPLEXITY_API_KEY="your_api_key_here"
In practice, set both to the same key.

Get API Key

Generate your Perplexity API key from the API portal.

​
Sonar Chat Completions

Call Perplexity’s Sonar models through litellm.completion with the perplexity/ model prefix: 
from litellm import completion
import os

os.environ["PERPLEXITYAI_API_KEY"] = "your_api_key_here"

response = completion(
    model="perplexity/sonar-pro",
    messages=[
        {"role": "user", "content": "What are the latest fusion breakthroughs?"}
    ],
)

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

​
Streaming

from litellm import completion

response = completion(
    model="perplexity/sonar-pro",
    messages=[{"role": "user", "content": "Explain quantum computing."}],
    stream=True,
)

for chunk in response:
    print(chunk)

​
Reasoning Effort

For reasoning-capable Sonar models, pass reasoning_effort to control depth: 
response = completion(
    model="perplexity/sonar-reasoning",
    messages=[{"role": "user", "content": "Walk through your reasoning."}],
    reasoning_effort="high",  # "low" | "medium" | "high"
)

​
Supported Sonar Models

ModelLiteLLM Identifier
sonarperplexity/sonar
sonar-properplexity/sonar-pro
sonar-reasoningperplexity/sonar-reasoning
sonar-reasoning-properplexity/sonar-reasoning-pro
sonar-deep-researchperplexity/sonar-deep-research

​
Agent API

Use litellm.responses to call the Agent API, which routes through Perplexity to third-party models with tool orchestration and presets.

​
Presets

from litellm import responses
import os

os.environ["PERPLEXITY_API_KEY"] = "your_api_key_here"

response = responses(
    model="perplexity/preset/pro-search",
    input="What are the latest developments in AI?",
    custom_llm_provider="perplexity",
)

print(response.output)
Available presets: fast-search, pro-search, deep-research, advanced-deep-research.

​
Tool Use (web_search and fetch_url)

from litellm import responses

response = responses(
    model="perplexity/openai/gpt-5.2",
    input="Research quantum computing breakthroughs and cite sources.",
    custom_llm_provider="perplexity",
    tools=[
        {"type": "web_search"},
        {"type": "fetch_url"},
    ],
    instructions="Use web_search and fetch_url to gather citations.",
    max_output_tokens=1000,
    temperature=0.7,
)

print(response.output)

​
Structured Outputs

from litellm import responses

response = responses(
    model="perplexity/preset/pro-search",
    input="Extract key facts about the Eiffel Tower.",
    custom_llm_provider="perplexity",
    text={
        "format": {
            "type": "json_schema",
            "name": "facts",
            "schema": {
                "type": "object",
                "properties": {
                    "name": {"type": "string"},
                    "height_meters": {"type": "number"},
                    "year_built": {"type": "integer"},
                },
                "required": ["name", "height_meters", "year_built"],
            },
            "strict": True,
        }
    },
)

​
Supported Third-Party Models via Agent API

ProviderModels
OpenAIperplexity/openai/gpt-5.5, perplexity/openai/gpt-5.4, perplexity/openai/gpt-5.4-mini, perplexity/openai/gpt-5.2, perplexity/openai/gpt-5.1, perplexity/openai/gpt-5-mini
Anthropicperplexity/anthropic/claude-opus-4-7, perplexity/anthropic/claude-opus-4-6, perplexity/anthropic/claude-sonnet-4-6, perplexity/anthropic/claude-opus-4-5, perplexity/anthropic/claude-sonnet-4-5, perplexity/anthropic/claude-haiku-4-5
Googleperplexity/google/gemini-3.1-pro-preview, perplexity/google/gemini-3-flash-preview, perplexity/google/gemini-3.1-flash-lite
xAIperplexity/xai/grok-4.20-non-reasoning
Perplexityperplexity/perplexity/sonar
See the Agent API model list for the canonical, up-to-date catalogue.

​
LiteLLM Proxy

Run LiteLLM as a self-hosted proxy that fronts Perplexity (and any other provider) behind a single OpenAI-compatible endpoint.

​
config.yaml

model_list:
  - model_name: perplexity-sonar-reasoning
    litellm_params:
      model: perplexity/sonar-reasoning
      api_key: os.environ/PERPLEXITYAI_API_KEY

  - model_name: perplexity-pro-search
    litellm_params:
      model: perplexity/preset/pro-search
      api_key: os.environ/PERPLEXITY_API_KEY

​
Start the Proxy

litellm --config /path/to/config.yaml

​
Call the Proxy

curl http://0.0.0.0:4000/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer anything" \
  -d '{
    "model": "perplexity-sonar-reasoning",
    "messages": [{"role": "user", "content": "Who won the World Cup in 2022?"}],
    "reasoning_effort": "high"
  }'

LiteLLM Perplexity Docs

Official LiteLLM Perplexity provider docs.

LiteLLM Docs

Full LiteLLM documentation.

Perplexity Agent API

Agent API reference and presets.

Perplexity Models

Available Sonar and Agent API models.

​
Support

Need help with the integration?