HELM VHELM - Crossmodal 3600 - Spanish Locale: leaderboard
VHELM (Stanford CRFM HELM for vision-language models, run v2.1.2) Crossmodal-3600 captioning of the 100 images from the Spanish locale: the model writes a short English caption, rated 1-5 by a Prometheus-Vision judge against human reference captions. Mean rating on the 1-5 scale, higher is better.
Metric: Prometheus Vision Rating (1-5). Source: crfm.stanford.edu. Saturation forecast: Estimated already saturated. 48 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | O4 Mini (2025-04-16) | 3.97 | #183 |
| 2 | GPT-4.1 Mini | 3.82 | #363 |
| 3 | GPT-4 Turbo | 3.8 | #417 |
| 4 | Claude 3.5 Sonnet (20240620) | 3.71 | #308 |
| 5 | Gemini 2.5 Pro (Preview 03-25) | 3.7 | #126 |
| 6 | Claude 3.5 Sonnet (20241022) | 3.65 | #302 |
| 7 | Gemini 1.5 Pro (002) | 3.57 | #368 |
| 8 | GPT-4.1 (2025-04-14) | 3.56 | #257 |
| 9 | O3 (2025-04-16) | 3.51 | #125 |
| 10 | O1 (2024-12-17) | 3.51 | #157 |
| 11 | Gemini 2.0 Pro (Preview 02-05) | 3.51 | #318 |
| 12 | GPT-4.5 | 3.49 | #278 |
| 13 | GPT-4o Mini (2024-07-18) | 3.47 | #568 |
| 14 | GPT-4.1 Nano | 3.39 | #730 |
| 15 | Claude 3 Opus (20240229) | 3.38 | #428 |
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-crossmodal-3600-spanish-locale · How It Works · Data refreshed daily, snapshot 2026-11-05.