HELM VHELM - Math Vista - College (Free Form): leaderboard
VHELM (Stanford CRFM HELM for vision-language models, run v2.1.2) MathVista items at grade level college in free form format: mathematical reasoning over plots, charts, diagrams and photos; 57 questions, exact match. Percent correct, 0-100, higher is better.
Metric: Exact Match (%). Source: crfm.stanford.edu. Saturation forecast: Estimated already saturated. 48 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | O3 (2025-04-16) | 57.89 | #125 |
| 2 | O4 Mini (2025-04-16) | 56.14 | #183 |
| 3 | Gemini 2.0 Flash | 50.88 | #351 |
| 4 | Gemini 2.5 Pro (Preview 03-25) | 50.88 | #126 |
| 5 | O1 (2024-12-17) | 50.88 | #157 |
| 6 | Gemini 2.0 Pro (Preview 02-05) | 49.12 | #318 |
| 7 | GPT-4.1 (2025-04-14) | 45.61 | #257 |
| 8 | Gemini 1.5 Pro (002) | 43.86 | #368 |
| 9 | Gemini 2.0 Flash (Preview) | 42.11 | #376 |
| 10 | Gemini 2.0 Flash Lite | 40.35 | #455 |
| 11 | GPT-4.5 | 38.6 | #278 |
| 12 | GPT-4.1 Mini | 36.84 | #363 |
| 13 | Claude 3.5 Sonnet (20241022) | 36.84 | #302 |
| 14 | GPT-4o (2024-08-06) | 35.09 | #343 |
| 15 | GPT-4o (2024-11-20) | 35.09 | #392 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=helm-vhelm-math-vista-college-free-form · How It Works · Data refreshed daily, snapshot 2026-11-05.