LiveBench Plot Unscrambling: leaderboard
LiveBench shuffles the sentences of a plot synopsis taken from a recently released film and asks the model to put them back in the original order. Grading compares the model's ordering with the true one and gives partial credit from the edit distance between the two sequences. Source material is refreshed to limit contamination, and higher is better. Source: arxiv.org/abs/2406.19314.
Metric: Score. Source: livebench.ai. Saturation forecast: Around May 2027. 71 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-6 Astra (Max) | 86.29 | #6 (GPT-6 Astra) |
| 2 | GPT-6.1 Sol (Max) | 84.38 | #7 (GPT-6.1 Sol) |
| 3 | GPT-6.1 Sol (xHigh) | 81.94 | #7 (GPT-6.1 Sol) |
| 4 | GPT-5.6 Sol (Max) | 79.05 | #19 (GPT-5.6 Sol) |
| 5 | Claude Fable 5 (Max) | 78.72 | #16 (Claude Fable 5) |
| 6 | GPT-6 Sol (Max) | 77.9 | #14 (GPT-6 Sol) |
| 7 | Claude Opus 5.5 (Max) | 77.48 | #3 (Claude Opus 5.5) |
| 8 | Claude Opus 5.5 (xHigh) | 77.16 | #3 (Claude Opus 5.5) |
| 9 | Claude Fable 5.1 (Max) | 75.17 | #9 (Claude Fable 5.1) |
| 10 | Muse Wonder (Max) | 74.8 | #15 (Muse Wonder) |
| 11 | Claude Opus 5 (Max) | 74.73 | #12 (Claude Opus 5) |
| 12 | Gemini 3.1 Pro (Preview) (High) | 74.13 | #58 (Gemini 3.1 Pro (Preview)) |
| 13 | GPT-5.5 (xHigh) | 74.09 | #31 (GPT-5.5) |
| 14 | Gemini 3.8 Flash (High) | 73.38 | #21 (Gemini 3.8 Flash) |
| 15 | Kimi K3 | 72.58 | #30 |
No result here: #8 Gemini 4 Argon, #10 Muse Wonder Full.
Interactive version: theaggregate.ai/benchmark?slug=livebench-plot-unscrambling · How It Works · Data refreshed daily, snapshot 2026-10-20.