perplexity: sonar
2 benchmarks
Leaderboard position
#665
Overall rank (of 1121 models)
740
Elo rating
2
Benchmarks
Top 59.3%
Percentile
In perplexity (of 7 models)
Above: inclusionAI: ui-venus-1.5-2b #664
Below: Tencent: hunyuanimage-3.0-fal #666
Benchmark highlights
Nearest competitor from another company
The model from another company with the most similar benchmark scores.
appeared as nearest in 1 of 2 benchmarks
1/2
50.0%
1/2
50.0%
| Model | Votes | % of benchmarks |
|---|---|---|
| StepFun: step-3-vl-10b | 1 | 50.0% |
| DeepSeek: deepseek-v3 | 1 | 50.0% |
Benchmarks
Predicted scores
Estimated from 25 similar models and the global Elo rating. These are projections, not measured results.
| Benchmark | Predicted | Confidence | Based on |
|---|---|---|---|
| MATH-500 | ~84.4 | 50% | — |
| AIME 2025 | ~62.35 | 50% | — |
| AIME 2026 | ~71.28 | 50% | — |
| HMMT Feb 2026 | ~62.78 | 50% | — |
| GPQA Diamond | ~55.57 | 50% | — |
| MMLU-Pro | ~58.41 | 50% | — |
| Humanity's Last Exam (no tools) | ~9.6 | 50% | — |
| IFEval | ~80.71 | 50% | — |
| IFBench | ~51.44 | 50% | — |
| HumanEval | ~70.39 | 50% | — |
| MBPP | ~72.69 | 50% | — |
| Terminal-Bench v2.1 | ~17.54 | 50% | — |
| MMLU | ~64.72 | 50% | — |
| SWE-bench Verified | ~35.6 | 50% | — |
| SWE-bench Pro | ~14.79 | 50% | — |
| SciCode | ~25.14 | 50% | — |
| AA-LCR | ~40.86 | 50% | — |
| Artificial Analysis Intelligence Index v5.0.1 | ~12 | 50% | — |
| Humanity's Last Exam (with tools) | ~13.13 | 50% | — |
| GSM8K | ~78.52 | 50% | — |
| -(CoderPPL) | ~-1.69 | 50% | — |
| LiveCodeBench v6 | ~53.32 | 50% | — |
| MMMU-Pro | ~50.8 | 50% | — |
| MATH | ~55.96 | 50% | — |
| BBH | ~62.79 | 50% | — |