ZEROBench-Sub: leaderboard
ZEROBench-Sub: the public subquestion subset of ZeroBench, extremely difficult visual reasoning puzzles over images; models answer the decomposed subquestions of each visual task.
Metric: Score (imported). Source: arxiv.org. Source: benchmarklist.com. Saturation forecast: Around January 2027. 41 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | O4 Mini | 29.05 | #180 |
| 2 | GPT-5 Mini (High) | 27.79 | #186 (GPT-5 Mini) |
| 3 | GPT-4.5 | 27.01 | #274 |
| 4 | GPT-5 (Medium) | 26.19 | #98 (GPT-5) |
| 5 | Gemini 2.5 Pro | 26.04 | #152 |
| 6 | O3 | 25.51 | #124 |
| 7 | Claude Opus 4.1 | 25.3 | #140 |
| 8 | Claude Opus 4 (Thinking) | 25.06 | #163 (Claude Opus 4) |
| 9 | Claude Sonnet 4 (Thinking) | 24.6 | #199 (Claude Sonnet 4) |
| 10 | Gemini 2.5 Flash | 24.48 | #245 |
| 11 | Llama 4 Maverick | 23.72 | #464 |
| 12 | Gemini 2.0 Flash | 23.24 | #342 |
| 13 | Claude Sonnet 4 | 23.01 | #199 |
| 14 | O1 Pro | 22.4 | #228 |
| 15 | Grok 4 | 21.58 | #181 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=zerobench-sub · How It Works · Data refreshed daily, snapshot 2026-10-20.