OpenVLM MathVista - LOG: leaderboard
OpenCompass OpenVLM MathVista logical reasoning subcategory score.
Metric: Accuracy (%). Source: huggingface.co. Source: huggingface.co. Saturation forecast: Around April 2027. 284 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 56.8 | #152 |
| 2 | GPT-4.1 (2025-04-14) | 54.1 | #250 |
| 3 | GPT-5 Mini (2025-08-07) | 54.1 | #172 |
| 4 | GPT-5 | 51.4 | #98 |
| 5 | GPT-4.1 Mini | 51.4 | #355 |
| 6 | GPT-5 Nano | 48.6 | #425 |
| 7 | Claude 3.5 Sonnet (20241022) | 45.9 | #291 |
| 8 | GPT-4o ChatGPT | 43.2 | |
| 9 | InternVL3-38B | 37.8 | #405 |
| 10 | Doubao-1.5-Pro | 37.8 | #287 |
| 11 | InternVL3-78B | 35.1 | #352 |
| 12 | Claude 3.5 Sonnet | 35.1 | #350 |
| 13 | Grok 2 (1212) | 35.1 | #496 |
| 14 | QVQ-72B-Preview | 35.1 | #440 |
| 15 | VLM-R1-3B-Math-0305 | 35.1 | #743 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=openvlm-mathvista-log · How It Works · Data refreshed daily, snapshot 2026-10-20.