HELM VHELM - Mm Star Perception - Fine-grained Perception: leaderboard
VHELM (Stanford CRFM HELM for vision-language models, run v2.1.2) MM-Star Fine-grained Perception subset: multiple-choice questions curated so that they cannot be answered without the image; 250 questions, prefix quasi-exact match. Percent correct, 0-100, higher is better.
Metric: Prefix Quasi-Exact Match (%). Source: crfm.stanford.edu. Saturation forecast: Estimated already saturated. 48 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro (Preview 03-25) | 74.8 | #126 |
| 2 | Llama 4 Maverick Instruct FP8 | 70.4 | #424 |
| 3 | Gemini 2.0 Flash | 68.4 | #351 |
| 4 | Gemini 2.0 Pro (Preview 02-05) | 68 | #318 |
| 5 | Gemini 2.0 Flash (Preview) | 67.6 | #376 |
| 6 | O1 (2024-12-17) | 66.4 | #157 |
| 7 | O3 (2025-04-16) | 66 | #125 |
| 8 | GPT-4.5 | 66 | #278 |
| 9 | O4 Mini (2025-04-16) | 65.2 | #183 |
| 10 | GPT-4.1 (2025-04-14) | 64.4 | #257 |
| 11 | Gemini 2.0 Flash Lite | 64 | #455 |
| 12 | Llama 4 Scout Instruct | 63.2 | #547 |
| 13 | GPT-4o (2024-05-13) | 63.2 | #337 |
| 14 | GPT-4o (2024-11-20) | 63.2 | #392 |
| 15 | Gemini 1.5 Pro (002) | 61.2 | #368 |
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-mm-star-perception-fine-grained-perception · How It Works · Data refreshed daily, snapshot 2026-11-05.