CLBench Life - Game Logs: leaderboard
CL-bench Life (2026) tests whether language models can learn from long, messy real-life context to solve human-written tasks, a task counting as solved only when every expert rubric item is met. This board is the task solving rate in percent on tasks over game logs such as match replays and hand histories; higher is better.
Metric: Task solving rate (%) on the 45 game-log (match replays, hand histories, gameplay commentary) tasks of CL-bench Life; a task is solved only when the response satisfies every human-written binary rubric item, judged by GPT-5.1 at high reasoning effort from the response and rubrics alone; reasoning mode at high effort where the model supports it, mean of three runs; higher is better. Source: arxiv.org. Saturation forecast: Around July 2027. 10 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | High | 34.1 | #99 |
| 2 | GPT-5.4 | High | 30.4 | #135 |
| 3 | Seed 2.0 Pro | High | 24.4 | #222 |
| 4 | Hy3-preview | High | 23.3 | #322 |
| 5 | Qwen 3.5 Plus | High | 20.7 | #242 |
| 6 | Claude Opus 4.6 | High | 17.8 | #113 |
| 7 | Kimi K2.5 | High | 17.8 | #266 |
| 8 | Grok 4.20 | High | 15.6 | #301 |
| 9 | DeepSeek V3.2 | Thinking | 12.6 | #419 |
| 10 | MiniMax-M2.5 | Thinking | 5.2 | #600 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=clbench-life-game-logs · How It Works · Data refreshed daily, snapshot 2026-11-21.