Sonar (Reasoning): benchmark results

Provider: Perplexity. Released 2025-02-11. Access: API.

Unified ELO 1545 ± 1, rank #1019 of 3078 rated models, from 9 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
AA GPQA Diamond62.32Accuracy (%)37.3
Artificial Analysis Intelligence Index8.67Intelligence Index36.9
DeepResearch Bench - Citation Accuracy52.58Score (%)33.3
DeepResearch Bench - Effective Citations13.37Effective Citations25
DeepResearch Bench - Insight32.59Score (%)13.6
DeepResearch Bench - Readability42.39Score (%)13.6
DeepResearch Bench37.75Score (%)9.1
DeepResearch Bench - Instruction Following44.42Score (%)9.1
DeepResearch Bench - Comprehensiveness34.73Score (%)6.8

Interactive version: theaggregate.ai/model?slug=sonar-reasoning · How It Works · Data refreshed daily, snapshot 2026-09-19.