Open LMM Reasoning - LogicVista - mechanical: leaderboard
Metric: Accuracy (%). Source: huggingface.co. 83 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4o ChatGPT | 75.7 |
| 2 | Claude 3.7 Sonnet | 68.9 |
| 3 | GPT-4.1 Mini | 67.6 |
| 4 | Claude 3.5 Sonnet (20241022) | 66.2 |
| 5 | InternVL3-38B | 66.2 |
| 6 | Doubao-1.5-Pro | 66.2 |
| 7 | Gemini 1.5 Pro (002) | 64.9 |
| 8 | Grok 2 (1212) | 64.9 |
| 9 | InternVL2.5-78B | 62.2 |
| 10 | QVQ-72B-Preview | 62.2 |
| 11 | GPT-4.1 (2025-04-14) | 60.8 |
| 12 | InternVL3-8B | 59.5 |
| 13 | InternVL3-78B | 56.8 |
| 14 | Gemini 2.0 Flash | 55.4 |
| 15 | Gemma 3 27B | 54.1 |
Interactive version: theaggregate.ai/benchmark?slug=open-lmm-reasoning-logicvista-mechanical · How It Works · Data refreshed daily, snapshot 2026-09-05.