Pencil Puzzle Bench - Hitori — leaderboard
Metric: Direct-ask Success Rate (%). Source: ppbench.com. 51 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.6 (Thinking) | 66.7 |
| 2 | Gemini 3.1 Pro (Preview) | 40 |
| 3 | Claude Opus 4.5 (Thinking) | 40 |
| 4 | GPT-5.2 (xHigh) | 33.3 |
| 5 | Claude Sonnet 4.6 (Thinking) | 33.3 |
| 6 | Grok 4.1 Fast | 26.7 |
| 7 | GPT-5.2 Pro | 26.7 |
| 8 | DeepSeek V3.2 | 20 |
| 9 | Gemini 3 Pro | 20 |
| 10 | GPT-5.2 (High) | 20 |
| 11 | GPT-5.1 (Medium) | 20 |
| 12 | Claude Sonnet 4.5 (Thinking) | 13.3 |
| 13 | Grok 4.1 Fast (Reasoning) | 13.3 |
| 14 | Kimi K2.5 | 13.3 |
| 15 | O3 | 13.3 |
Interactive version: theaggregate.ai/benchmark?slug=pencil-puzzle-bench-hitori · How the rankings work · Data refreshed daily, snapshot 2026-07-22.