OpenVLM MathVista - TQA: leaderboard
OpenCompass OpenVLM MathVista textbook question answering subcategory score.
Metric: Accuracy (%). Source: huggingface.co. Source: huggingface.co. Saturation forecast: Around January 2027. 284 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 84.8 | #98 |
| 2 | Gemini 2.5 Pro | 83.5 | #152 |
| 3 | GPT-5 Nano | 80.4 | #425 |
| 4 | GPT-5 Mini (2025-08-07) | 80.4 | #172 |
| 5 | Doubao-1.5-Pro | 79.1 | #287 |
| 6 | GPT-4.1 Mini | 77.8 | #355 |
| 7 | Gemini 2.0 Flash | 77.8 | #342 |
| 8 | GPT-4o ChatGPT | 77.2 | |
| 9 | GPT-4.1 (2025-04-14) | 76.6 | #250 |
| 10 | InternVL2.5-78B | 75.9 | #431 |
| 11 | Gemini 1.5 Pro (002) | 75.3 | #349 |
| 12 | Gemini 1.5 Flash (002) | 74.7 | #583 |
| 13 | Claude 3.7 Sonnet | 74.7 | #249 |
| 14 | GPT-4.5 | 74.7 | #274 |
| 15 | InternVL2.5-78B-CoT | 74.7 | #469 |
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-tqa · How It Works · Data refreshed daily, snapshot 2026-10-20.