HELM VHELM - Blink Reasoning: leaderboard
The "Blink Reasoning" 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 | Claude 3 Sonnet (20240229) | 82 | #681 |
| 2 | Claude 3 Opus (20240229) | 81.33 | #413 |
| 3 | Claude 3 Haiku (20240307) | 80.44 | #777 |
| 4 | GPT-4o Mini (2024-07-18) | 79.94 | #561 |
| 5 | Claude 3.5 Sonnet (20240620) | 79.78 | #298 |
| 6 | Llama 4 Maverick Instruct FP8 | 79.78 | #412 |
| 7 | GPT-4 Turbo | 79.44 | #406 |
| 8 | Gemini 2.0 Flash Lite | 79.11 | #453 |
| 9 | GPT-4.1 (2025-04-14) | 76.86 | #250 |
| 10 | O1 (2024-12-17) | 76.83 | #152 |
| 11 | Gemini 2.5 Pro (Preview 03-25) | 75.15 | #124 |
| 12 | O3 (2025-04-16) | 70.58 | #123 |
| 13 | GPT-4o (2024-05-13) | 69.42 | #326 |
| 14 | GPT-4.1 Mini | 68.97 | #351 |
| 15 | O4 Mini (2025-04-16) | 68.33 | #181 |
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-blink-reasoning · How It Works · Data refreshed daily, snapshot 2026-10-23.