Open LMM Reasoning - WeMath - Three-step(S3) — leaderboard
Metric: Accuracy (%). Source: huggingface.co. 83 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Doubao-1.5-Pro | 70.9 |
| 2 | GPT-4.1 (2025-04-14) | 61.8 |
| 3 | Gemini 2.0 Flash | 60.6 |
| 4 | GPT-4o ChatGPT | 58.8 |
| 5 | Claude 3.7 Sonnet | 56.4 |
| 6 | Claude 3.5 Sonnet (20241022) | 54.5 |
| 7 | Gemini 1.5 Pro (002) | 53.9 |
| 8 | QVQ-72B-Preview | 53.9 |
| 9 | InternVL3-38B | 52.7 |
| 10 | InternVL3-78B | 50.3 |
| 11 | InternVL3-14B | 49.7 |
| 12 | GPT-4.1 Nano | 49.1 |
| 13 | Grok 2 (1212) | 47.3 |
| 14 | InternVL3-8B | 46.7 |
| 15 | InternVL2.5-78B | 46.1 |
Interactive version: theaggregate.ai/benchmark?slug=open-lmm-reasoning-wemath-three-step-s3 · How the rankings work · Data refreshed daily, snapshot 2026-07-22.