Perplexity: Sonar Reasoning

Text input Text output
Author's Description

Sonar Reasoning is a reasoning model provided by Perplexity based on [DeepSeek R1](/deepseek/deepseek-r1). It allows developers to utilize long chain of thought with built-in web search. Sonar Reasoning is uncensored and hosted in US datacenters.

Key Specifications
Cost
$$$$$
Context
127K
Released
Jan 28, 2025
Speed
Ability
Reliability
Supported Parameters

This model supports the following parameters:

Include Reasoning Max Tokens Top P Frequency Penalty Reasoning Temperature Presence Penalty
Features

This model supports the following features:

Reasoning
Performance Summary

Perplexity's Sonar Reasoning model, based on DeepSeek R1, demonstrates exceptional reliability with a 98% success rate, consistently providing usable responses. However, it tends to have longer response times, ranking in the 18th percentile for speed, and is positioned at premium pricing levels, ranking in the 9th percentile for cost. The model exhibits a key strength in its ability to acknowledge uncertainty, achieving perfect 100% accuracy in the Hallucinations (Baseline) benchmark, indicating no hallucination. It also performs very strongly in Reasoning (98% accuracy, 92nd percentile) and Email Classification (99% accuracy, 72nd percentile). Its General Knowledge (97% accuracy) and Mathematics (89% accuracy) capabilities are solid. A notable weakness is observed in Instruction Following, where it achieved only 34.7% accuracy, placing it in the 35th percentile. Coding performance is moderate at 81% accuracy, and Ethics performance is 95% accurate but ranks in the lower 28th percentile. Its built-in web search and uncensored nature are additional features.

Model Pricing

Current Pricing

Feature Price (per 1M tokens)
Prompt $1
Completion $5
Request $0.005 per request

Price History

Available Endpoints
Provider Endpoint Name Context Length Pricing (Input) Pricing (Output)
Perplexity
Perplexity | perplexity/sonar-reasoning 127K $1 / 1M tokens $5 / 1M tokens
Benchmark Results
Benchmark Category Reasoning Strategy Free Executions Accuracy Cost Duration
Other Models by perplexity