LLM Stats (ZEROBench): leaderboard
Zero-shot visual reasoning benchmark with tasks designed so that no current model achieves high accuracy, measuring frontier visual capabilities.
Metric: Score (%). Source: llm-stats.com. Saturation forecast: Around January 2027. 15 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | DeepSeek V4.1 Flash (1017) | 51 | #20 |
| 2 | Muse Spark Open | 50 | #54 |
| 3 | Seed 2.1 Pro | 18 | #67 |
| 4 | Kimi K2.5 | 11 | #144 |
| 5 | Seed 1.8 | 11 | #139 |
| 6 | Seed 2.1 Turbo | 11 | #86 |
| 7 | Qwen 3.5 27B | 10 | #187 |
| 8 | Qwen 3.5 122B A10B | 9 | #173 |
| 9 | Qwen 3.5 35B A3B | 8 | #256 |
| 10 | Seed 2.0 Mini | 7 | #276 |
| 11 | Qwen 3 VL 235B A22B (Thinking) | 4 | #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 · How It Works · Data refreshed daily, snapshot 2026-10-20.