Fiction.LiveBench: leaderboard
Long-context comprehension benchmark using 36 questions across 30 stories at varying lengths. Tests deep narrative understanding, subtext, and plot hole detection - not just retrieval.
Metric: Accuracy (%). Source: fiction.live. Saturation forecast: Estimated already saturated. 41 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | O3 (2025-04-16) (Medium) | 100 | #121 (O3 (2025-04-16)) |
| 2 | Grok 4 | 96.9 | #181 |
| 3 | GPT-5 (Medium) | 96.9 | #98 (GPT-5) |
| 4 | Gemini 2.5 Pro (Preview 03-25) | 90.6 | #123 |
| 5 | Gemini 2.5 Pro (Preview 06-05) | 87.5 | #106 |
| 6 | Kimi K2.5 | 78.1 | #144 |
| 7 | Grok 4 Fast | 75 | #254 |
| 8 | Gemini 2.5 Pro (Preview 05-06) | 71.9 | #147 |
| 9 | Qwen 3 235B A22B 2507 (Thinking) | 68.8 | #260 (Qwen 3 235B A22B 2507) |
| 10 | Gemini 2.5 Flash (Preview 05-20) | 68.8 | #216 |
| 11 | GPT-4.5 (Preview) | 63.9 | #244 |
| 12 | GPT-4.1 | 62.5 | #248 |
| 13 | Qwen 3 Next 80B A3B Instruct | 62.5 | #368 |
| 14 | Gemini 2.0 Flash (001) | 62.5 | #360 |
| 15 | O4 Mini (Medium) | 62.5 | #180 (O4 Mini) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=fiction-livebench · How It Works · Data refreshed daily, snapshot 2026-10-20.