LiveBench Typos: leaderboard
LiveBench injects common human typos into a recent arXiv abstract and asks the model to return the passage with the misspellings fixed and nothing else altered. Scoring is all or nothing: the answer counts only if the exact corrected text appears inside it, so any other edit loses the point. The board is the percent of passages scored correct, higher is better. Source: arxiv.org/abs/2406.19314.
Metric: Score. Source: livebench.ai. Saturation forecast: Around January 2027. 71 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Fable 5.1 (Max) | 94 | #9 (Claude Fable 5.1) |
| 2 | Claude Fable 5 (Max) | 94 | #16 (Claude Fable 5) |
| 3 | Claude Opus 5 (Max) | 92 | #12 (Claude Opus 5) |
| 4 | Gemini 3.8 Flash (High) | 90 | #21 (Gemini 3.8 Flash) |
| 5 | Gemini 3.7 Flash (High) | 90 | #23 (Gemini 3.7 Flash) |
| 6 | GPT-5.5 (xHigh) | 88 | #31 (GPT-5.5) |
| 7 | Gemini 3.6 Flash (High) | 88 | #51 (Gemini 3.6 Flash) |
| 8 | Qwen 3.8 Max | 86 | #35 |
| 9 | Gemini 3.5 Flash (High) | 86 | #60 (Gemini 3.5 Flash) |
| 10 | GPT-5.6 Terra (Max) | 86 | #41 (GPT-5.6 Terra) |
| 11 | GPT-6.1 Sol (Max) | 86 | #7 (GPT-6.1 Sol) |
| 12 | Muse Spark 1.3 (xHigh) | 86 | #24 (Muse Spark 1.3) |
| 13 | Claude Sonnet 5.5 (Max) | 86 | #11 (Claude Sonnet 5.5) |
| 14 | MiniMax-M3 | 84 | #136 |
| 15 | Kimi K3 | 84 | #30 |
No result here: #8 Gemini 4 Argon, #10 Muse Wonder Full.
Interactive version: theaggregate.ai/benchmark?slug=livebench-typos · How It Works · Data refreshed daily, snapshot 2026-10-20.