A high-performance reasoning model leveraging advanced multi-step CoT reasoning and enhanced information retrieval.

Model Type: Reasoning
Use Case: Best suited for dealing with complex topics that require advanced multi-step reasoning.

Key Features:

  • Enhanced Chain-of-thought (CoT) reasoning
  • 2x more citations on average than Sonar Reasoning
  • Uses advanced information retrieval architecture

Real-World Examples:

  • Performing competitive analyses for new products
  • Understanding and exploring complex scientific topics
  • Making detailed travel plans

​
Pricing

Legacy Pricing (Active Until 04/18/2025)

Legacy Pricing (Default if no search mode is specified)

Pricing:

Pricing ComponentCost
Input Tokens (Per Million)$2
Output Tokens (Per Million)$8
Price per 1,000 Requests$5

New Pricing (Available Now)

New Pricing with Search Modes
Explicit API call with search_context_size required to use new pricing.

MetricHigh  Medium  Low  
Input Tokens (Per Million)$2$2$2
Output Tokens (Per Million)$8$8$8
Price per 1,000 Requests$14$10$6

​
Test the Model

curl --request POST \
  --url https://api.perplexity.ai/chat/completions \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '{
  "model": "sonar-reasoning-pro",
  "messages": [
    {"role": "user", "content": "Analyze the feasibility of fusion energy becoming a mainstream power source by 2040."}
  ],
  "max_tokens": 300
}'
  • The sonar-reasoning-pro model is designed to output a <think> section containing reasoning tokens, immediately followed by a valid JSON object. As a result, the response_format parameter does not remove these reasoning tokens from the output. We recommend using a custom parser to extract the valid JSON portion, and an example implementation can be found here.