NonoBench (v1.0): leaderboard
Metric: Standard accuracy (%), correct / non-failed attempts; version 1.0. Source: www.nonobench.com. Saturation forecast: Around January 2027. 38 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.2 (High) | 66.67 |
| 2 | Claude Opus 4.5 (High) | 60 |
| 3 | Gemini 3 Pro (Preview) (High) | 56.67 |
| 4 | GPT-5.2 (xHigh) | 56.67 |
| 5 | Gemini 3 Flash (Preview) (High) | 50 |
| 6 | GPT-5.2 (Low) | 50 |
| 7 | GPT-OSS-120B (High) | 43.33 |
| 8 | DeepSeek V3.2 (High) | 43.33 |
| 9 | Kimi K2.5 (High) | 43.33 |
| 10 | DeepSeek V3.2 Speciale | 40 |
| 11 | Grok 4 | 33.33 |
| 12 | Qwen 3 Next 80B A3B (Thinking) | 33.33 |
| 13 | GPT-OSS-120B (Low) | 33.33 |
| 14 | Claude Sonnet 4.5 (Thinking) | 30 |
| 15 | Kimi K2 (Thinking) | 26.67 |
Interactive version: theaggregate.ai/benchmark?slug=nonobench-v1-0 · How It Works · Data refreshed daily, snapshot 2026-10-09.