FrontierCode Extended: leaderboard
Best published five-trial mean rubric score for each model across tested reasoning efforts on 150 Extended tasks; selected effort and evaluation harness are retained.
Metric: Rubric Score (%). Source: cognition.ai. Saturation forecast: Around November 2027. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 (High) | 44.76 | #31 (GPT-5.5) |
| 2 | Kimi K2.6 | 37.01 | #104 |
| 3 | GPT-5.4 Mini (xHigh) | 36.01 | #212 (GPT-5.4 Mini) |
| 4 | Kimi K2.5 | 22.69 | #144 |
| 5 | MiniMax-M2.7 | 19.86 | #258 |
| 6 | MiniMax-M2.5 | 15.78 | #303 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=frontiercode-extended · How It Works · Data refreshed daily, snapshot 2026-10-20.