P

Perplexity: Sonar Pro

Perplexity
Vision
200K
Context
$3.00
Input /1M
$15.00
Output /1M
8K
Max Output
Run API Call Open in Chat

Perplexity: Sonar Pro

Overview

Property Value
Model Name Perplexity: Sonar Pro
Model ID perplexity/sonar-pro
Provider Perplexity
Created March 7, 2025
Context Length 200,000 tokens
Max Completion Tokens 8,000

Description

Perplexity Sonar Pro is an advanced search-augmented language model designed for in-depth, multi-step queries with added extensibility. It offers approximately double the citations per search compared to standard Sonar, making it ideal for research and complex information retrieval tasks.

The model features a larger context window that supports longer and more nuanced searches, enabling sophisticated follow-up questions and multi-turn conversations that maintain context across complex research workflows.

Pricing

Category Cost
Input $3.00 per 1M tokens ($0.000003/token)
Output $15.00 per 1M tokens ($0.000015/token)
Web Search $5.00 per thousand searches ($0.005/search)

Note: Pricing includes Perplexity search costs. Web search pricing may vary based on search depth thresholds (low/medium/high).

Supported Parameters

Parameter Description
max_tokens Maximum number of tokens to generate
temperature Controls randomness in output (0-2)
top_p Nucleus sampling threshold
top_k Top-k sampling parameter
frequency_penalty Reduces repetition of frequent tokens
presence_penalty Reduces repetition of present tokens
web_search_options Configuration for web search behavior

Use Cases

Best For

  • Research Tasks: Academic research, literature reviews, fact-checking
  • Complex Queries: Multi-step questions requiring extensive web search
  • Citation-Heavy Work: Tasks requiring verified sources and citations
  • Long-Form Analysis: Extended conversations with context preservation
  • Image Analysis: Understanding and analyzing uploaded images
  • Simple Q&A: Basic questions that don't require web search
  • Creative Writing: Tasks focused on pure generation without factual grounding
  • Real-time Applications: Where search latency is a concern
Model Description
perplexity/sonar Standard Sonar model with basic search capabilities
perplexity/sonar-deep-research Extended research capabilities for complex queries
perplexity/sonar-reasoning Version with explicit reasoning support
perplexity/sonar-reasoning-pro Pro version with reasoning capabilities

Modalities

Type Support
Input Text, Image
Output Text

Features & Capabilities

Supported Features

  • Multipart Support: Enabled for handling complex multi-part requests
  • Tool Choice: Supports none, auto, required, and function types
  • Web Search Integration: Built-in web search with configurable depth thresholds
  • Image Input: Can process and understand image content
  • Large Context Window: 200K tokens for extended conversations

Limitations

  • No Reasoning Support: Does not support chain-of-thought or explicit reasoning modes
  • Text Output Only: Cannot generate images or other media formats

Data Policy

Policy Status
Training Usage No - Data not used for training
Prompt Retention No - Prompts are not retained
Publication Cannot be published

Web Search Options

The model supports variable web search pricing thresholds:

Threshold Description
Low Basic web search with minimal citations
Medium Balanced search depth with moderate citations
High Deep search with maximum citations (~2x standard)

Usage Examples

OpenAI-Compatible API

curl -X POST https://api.langmart.ai/v1/chat/completions \
  -H "Authorization: Bearer $LANGMART_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "perplexity/sonar-pro",
    "messages": [
      {
        "role": "user",
        "content": "What are the latest developments in quantum computing?"
      }
    ],
    "max_tokens": 4096,
    "temperature": 0.7
  }'

With Web Search Options

curl -X POST https://api.langmart.ai/v1/chat/completions \
  -H "Authorization: Bearer $LANGMART_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "perplexity/sonar-pro",
    "messages": [
      {
        "role": "user",
        "content": "Compare the latest AI safety research papers from 2024"
      }
    ],
    "web_search_options": {
      "search_depth": "high"
    }
  }'

Version History

Date Version Notes
March 7, 2025 Initial Release Launch of Sonar Pro with enhanced search capabilities

Last Updated: December 2025 Source: LangMart Model Registry