HELM VHELM - Vibe Eval: leaderboard
The "Vibe Eval" subtask of HELM VHELM. Stanford HELM's VHELM project, the holistic evaluation of vision-language models: image-grounded question answering, OCR, chart and diagram reading, spatial reasoning, robustness, bias, fairness and toxicity scenarios scored under one ha...
Metric: Prometheus Vision rating. 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) | 3.38 | #124 |
| 2 | O4 Mini (2025-04-16) | 3.1 | #181 |
| 3 | O3 (2025-04-16) | 2.93 | #123 |
| 4 | Claude 3.5 Sonnet (20241022) | 2.85 | #291 |
| 5 | GPT-4.1 (2025-04-14) | 2.82 | #250 |
| 6 | Claude 3.5 Sonnet (20240620) | 2.78 | #298 |
| 7 | O1 (2024-12-17) | 2.77 | #152 |
| 8 | Claude 3 Opus (20240229) | 2.74 | #413 |
| 9 | Llama 4 Maverick Instruct FP8 | 2.69 | #412 |
| 10 | GPT-4o (2024-11-20) | 2.69 | #378 |
| 11 | GPT-4o (2024-05-13) | 2.68 | #326 |
| 12 | GPT-4.1 Mini | 2.67 | #351 |
| 13 | Gemini 2.0 Flash (Preview) | 2.66 | #366 |
| 14 | Claude 3 Sonnet (20240229) | 2.63 | #681 |
| 15 | GPT-4 Turbo | 2.62 | #406 |
No result here: #4 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=helm-vhelm-vibe-eval · How It Works · Data refreshed daily, snapshot 2026-10-23.