LLM Stats (ZEROBench-Sub): leaderboard
Subset of ZEROBench focusing on zero-shot visual reasoning tasks designed to challenge multimodal models on complex visual understanding and deduction.
Metric: Score (%). Source: llm-stats.com. Saturation forecast: Around August 2027. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.5 27B | 36.2 | #187 |
| 2 | Qwen 3.5 122B A10B | 36.2 | #173 |
| 3 | Seed 2.0 Mini | 36.2 | #276 |
| 4 | Qwen 3.6 35B A3B | 34.4 | #226 |
| 5 | Qwen 3.5 35B A3B | 34.1 | #256 |
| 6 | Qwen 3 VL 235B A22B (Thinking) | 27.7 | #238 (Qwen 3 VL 235B A22B) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=llm-stats-zerobench-sub · How It Works · Data refreshed daily, snapshot 2026-10-20.