HELM VHELM - Knowledge: leaderboard
The "Knowledge" 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: Mean win rate. 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) | 98.72 | #124 |
| 2 | GPT-4.1 (2025-04-14) | 83.4 | #250 |
| 3 | O4 Mini (2025-04-16) | 82.13 | #181 |
| 4 | O3 (2025-04-16) | 81.28 | #123 |
| 5 | Gemini 2.0 Pro (Preview 02-05) | 78.09 | #305 |
| 6 | GPT-4.5 | 76.17 | #276 |
| 7 | GPT-4o (2024-05-13) | 72.77 | #326 |
| 8 | Gemini 2.0 Flash (Preview) | 72.55 | #366 |
| 9 | Gemini 2.0 Flash | 70.21 | #344 |
| 10 | GPT-4o (2024-08-06) | 69.79 | #335 |
| 11 | O1 (2024-12-17) | 69.36 | #152 |
| 12 | GPT-4.1 Mini | 68.94 | #351 |
| 13 | Gemini 1.5 Pro (002) | 67.87 | #350 |
| 14 | GPT-4o (2024-11-20) | 66.81 | #378 |
| 15 | Llama 4 Maverick Instruct FP8 | 59.15 | #412 |
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-knowledge · How It Works · Data refreshed daily, snapshot 2026-10-23.