SAHARA - Mathematical Word Problems: leaderboard
Metric: Exact match (%). Source: huggingface.co. 47 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 82.79 |
| 2 | Gemini 2.5 Pro | 80.7 |
| 3 | GPT-5 | 74.43 |
| 4 | Gemini 2.5 Flash | 72.21 |
| 5 | GPT-5.1 | 70.02 |
| 6 | GPT-5.6 Luna | 53.89 |
| 7 | Claude Sonnet 4 (20250514) | 45.9 |
| 8 | Claude Sonnet 4 | 42.82 |
| 9 | GPT-4.1 | 29.5 |
| 10 | Qwen 3.6 27B | 29.41 |
| 11 | Qwen 3.5 27B | 28.01 |
| 12 | Qwen 3.6 35B A3B | 22.6 |
| 13 | Gemma 4 31B (IT) | 22 |
| 14 | Llama 4 Scout Instruct | 19.81 |
| 15 | c4ai-command-a-03-2025 | 11.1 |
Interactive version: theaggregate.ai/benchmark?slug=sahara-mathematical-word-problems · How It Works · Data refreshed daily, snapshot 2026-09-19.