OpenVLM MathVista - ALG: leaderboard
OpenCompass OpenVLM MathVista algebraic reasoning subcategory score.
Metric: Accuracy (%). Source: huggingface.co. Source: huggingface.co. Saturation forecast: Estimated already saturated. 284 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 93.6 | #98 |
| 2 | GPT-5 Nano | 89.7 | #425 |
| 3 | GPT-5 Mini (2025-08-07) | 89.7 | #172 |
| 4 | Doubao-1.5-Pro | 89.7 | #287 |
| 5 | Gemini 2.5 Pro | 87.5 | #152 |
| 6 | Gemini 2.0 Flash | 83.3 | #342 |
| 7 | InternVL3-78B | 82.6 | #352 |
| 8 | Gemini 1.5 Pro (002) | 80.1 | #349 |
| 9 | InternVL3-38B | 78.3 | #405 |
| 10 | GPT-4.5 | 76.9 | #274 |
| 11 | InternVL3-14B | 76.5 | #505 |
| 12 | Gemma 3 27B | 76.5 | #607 |
| 13 | QVQ-72B-Preview | 75.4 | #440 |
| 14 | InternVL3-8B | 75.1 | #620 |
| 15 | GPT-4o ChatGPT | 75.1 |
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-alg · How It Works · Data refreshed daily, snapshot 2026-10-20.