ZeroBench (Main questions, pass@5)
Multimodal · 2025-02-13
This row reports pass@5 on ZeroBench's 100 main visual-reasoning questions: the percentage answered correctly at least once across five sampled attempts.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-6 Astra | 52.0 |
| Claude Opus 5.5 | 43.0 |
| GPT-6 Sol | 41.0 |
| Kimi K3.1 Pro | 39.0 |
| Claude Opus 5 | 26.0 |
| Claude Fable 5 | 24.0 |
| GPT-5.4 | 23.0 |
| Kimi K3 | 23.0 |
| Gemini 3.5 Flash | 22.0 |
| GPT-5.6 Luna | 21.0 |
| GPT-6 Luna | 20.0 |
| Gemini 3 Pro Preview | 20.0 |
| GPT-5.6 Terra | 19.0 |
| Gemini 3.1 Pro Preview | 19.0 |
| Gemini 3.6 Flash | 18.0 |
| Opus 4.8 | 18.0 |
| GPT-5.2 | 18.0 |
| Grok 4.6 | 17.0 |
| Grok 4.5 | 15.0 |
| Opus 4.7 | 15.0 |
| Gemini 3 Flash Preview | 15.0 |
| Gemini 3.5 Flash-Lite | 14.0 |
| Sonnet 5 | 13.0 |
| Grok 4.3 | 13.0 |
| Sonnet 4.6 | 12.0 |
| Opus 4.6 | 12.0 |
| GPT-5.4 Mini | 11.0 |
| O4 Mini | 11.0 |
| Opus 4.5 | 9.0 |
| GPT-5 Mini | 9.0 |
| Opus 4.1 | 8.0 |
| GPT-5 | 7.0 |
| GPT-5.1 | 6.0 |
| Grok 4 | 6.0 |
| Opus 4 | 6.0 |
| Sonnet 4 | 6.0 |
| o3 | 6.0 |
| GPT-4.5 | 6.0 |
| Claude 3.7 Sonnet | 6.0 |
| Gemini 3.1 Flash-Lite Preview | 5.0 |