LisanBench: leaderboard
LisanBench evaluates LLMs on open-ended word-chain generation, stressing vocabulary, planning, recall, persistence, and exact edit-distance rule following.
Metric: Mean Path Length / Current Maximum. Source: lisanbench.com. Status: saturation imminent. 154 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 5 (High) | 1 |
| 2 | Claude Opus 4.7 (xHigh) | 0.93 |
| 3 | Claude Fable 5 (Medium) | 0.89 |
| 4 | Kimi K3 (Max) | 0.68 |
| 5 | GPT-5.5 (Medium) | 0.64 |
| 6 | Claude Opus 4.8 (High) | 0.61 |
| 7 | Claude Opus 5 (Medium) | 0.53 |
| 8 | Gemini 3.1 Pro (Preview) (High) | 0.42 |
| 9 | Claude Sonnet 5 (High) | 0.37 |
| 10 | GPT-5.4 (Medium) | 0.35 |
| 11 | GPT-5.6 Sol (Medium) | 0.3 |
| 12 | Gemini 3 Pro (Preview) (High) | 0.3 |
| 13 | Gemini 3.5 Flash (High) | 0.27 |
| 14 | GPT-5.2 (Medium) | 0.23 |
| 15 | Gemini 3.1 Pro (Preview) (Low) | 0.23 |
Interactive version: theaggregate.ai/benchmark?slug=lisanbench · How It Works · Data refreshed daily, snapshot 2026-09-05.