Sonar — benchmark results
Perplexity's search-grounded API model built on Llama 3.3 70B and served on Cerebras hardware for fast, cited real-time answers (February 2025). Provider: Perplexity. Released 2025-02-01. Access: API.
Unified ELO 1498 ± 60, rank #807 of 1776 rated models, from 18 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AA Humanity's Last Exam | 7.27 | Accuracy (%) | 52.3 |
| AA MATH-500 | 81.67 | Accuracy (%) | 48.5 |
| AI for Education Pedagogy - Technology | 77.36 | Accuracy (%) | 47.7 |
| Gapminder AI Worldview | 78.4 | Correct Rate (%) | 44.1 |
| AI for Education Pedagogy - Social studies | 71.82 | Accuracy (%) | 40 |
| AI for Education Pedagogy - Primary | 76.53 | Accuracy (%) | 35.9 |
| AA LiveCodeBench | 29.52 | Pass@1 (%) | 34.8 |
| Artificial Analysis Intelligence Index | 9.51 | Intelligence Index | 33 |
| AA MMLU-Pro | 68.92 | Accuracy (%) | 32.8 |
| AI for Education Pedagogy | 70.75 | Accuracy (%) | 30.5 |
| AI for Education Pedagogy - Secondary | 69.34 | Accuracy (%) | 29.5 |
| AI for Education Pedagogy - Science | 66.67 | Accuracy (%) | 25.7 |
Interactive version: theaggregate.ai/model?slug=sonar · How the rankings work · Data refreshed daily, snapshot 2026-07-22.