HELM VHELM - Seed Bench: leaderboard
The "Seed Bench" 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: PEM. Source: crfm.stanford.edu. Saturation forecast: Estimated already saturated. 48 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | O3 (2025-04-16) | 84.99 | #123 |
| 2 | GPT-4.5 | 84.69 | #276 |
| 3 | Gemini 2.5 Pro (Preview 03-25) | 83.96 | #124 |
| 4 | Gemini 1.5 Flash (002) | 83.75 | #585 |
| 5 | GPT-4.1 Mini | 83.57 | #351 |
| 6 | O1 (2024-12-17) | 83.48 | #152 |
| 7 | GPT-4 Turbo | 82.9 | #406 |
| 8 | GPT-4o (2024-05-13) | 82.78 | #326 |
| 9 | GPT-4o (2024-11-20) | 82.17 | #378 |
| 10 | GPT-4o (2024-08-06) | 82.05 | #335 |
| 11 | O4 Mini (2025-04-16) | 81.81 | #181 |
| 12 | Claude 3.5 Sonnet (20241022) | 81.54 | #291 |
| 13 | GPT-4.1 (2025-04-14) | 81.14 | #250 |
| 14 | GPT-4o Mini (2024-07-18) | 80.87 | #561 |
| 15 | Gemini 1.5 Pro (002) | 80.63 | #350 |
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-seed-bench · How It Works · Data refreshed daily, snapshot 2026-10-23.