MirrorCode: leaderboard
Epoch AI coding benchmark where models reimplement whole open-source programs from observed behavior rather than copying the original source. Reports per-target mean solve rates across program-reconstruction tasks.
Metric: Solve@100 Rate (%). Source: epoch.ai. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Fable 5 (High) | 63.9 | #16 (Claude Fable 5) |
| 2 | Claude Opus 4.7 (High) | 31.1 | #49 (Claude Opus 4.7) |
| 3 | GPT-5.6 Sol (High) | 20 | #19 (GPT-5.6 Sol) |
| 4 | GPT-5.4 (High) | 15.6 | #79 (GPT-5.4) |
| 5 | GPT-5.5 (High) | 10 | #31 (GPT-5.5) |
| 6 | Gemini 3.1 Pro (Preview) | 8.9 | #58 |
Interactive version: theaggregate.ai/benchmark?slug=mirrorcode · How It Works · Data refreshed daily, snapshot 2026-10-20.