MirrorCode
Reimplementing programs from their observed behaviour. Run by Epoch AI.
| Category | Coding |
|---|---|
| Metric | Accuracy |
| Source | Epoch AI (evaluator) |
| License | CC BY 4.0 |
| Benchmark page | epoch.ai |
| Status | Scored |
| Weight | 0.8 within Coding |
| Fit | Discrimination 2.35, noise 0.66 |
Results
| # | Model | Provider | Result | Date |
|---|---|---|---|---|
| 1 | Claude Sonnet 5.5 Max | Anthropic | 97.8% | 2026-10-05 |
| 2 | Claude Opus 5.5 Max | Anthropic | 95.6% | 2026-10-30 |
| 3 | Claude Fable 5.1 Max | Anthropic | 93.3% | 2026-09-09 |
| 4 | Claude Opus 5 Max | Anthropic | 87.8% | 2026-10-20 |
| 5 | Claude Haiku 5.5 Max | Anthropic | 78.9% | 2026-10-21 |
| 6 | Claude Opus 5.5 High | Anthropic | 78.9% | 2026-09-22 |
| 7 | GPT-6.1 Terra Max | OpenAI | 76.7% | 2026-10-30 |
| 8 | Claude Fable 5.1 High | Anthropic | 73.3% | 2026-09-10 |
| 9 | Claude Fable 5 Max | Anthropic | 70.0% | 2026-10-30 |
| 10 | Claude Opus 4.8 Max | Anthropic | 65.6% | 2026-10-21 |
| 11 | DeepSeek V4.1 Flash Max | DeepSeek | 64.4% | 2026-10-21 |
| 12 | Claude Fable 5 High | Anthropic | 63.9% | 2026-08-10 |
| 13 | Claude Sonnet 5 Max | Anthropic | 63.3% | 2026-10-21 |
| 14 | GPT-6.1 Sol Max | OpenAI | 61.1% | 2026-10-05 |
| 15 | GPT-6 Astra Max | OpenAI | 60.0% | 2026-10-29 |
| 16 | Gemini 4 Argon High | 58.9% | 2026-10-14 | |
| 17 | Grok 4.7 Extra High | xAI | 54.4% | 2026-10-29 |
| 18 | Kimi K3 Max | Moonshot AI | 54.4% | 2026-09-03 |
| 19 | Muse Spark 1.3 Max | Meta | 54.4% | 2026-10-20 |
| 20 | Qwen3.8 Max Extra High | Alibaba | 52.2% | 2026-10-21 |
| 21 | GPT-6 Sol Max | OpenAI | 48.9% | 2026-10-29 |
| 22 | Qwen3.8 Max 0902 Extra High | Alibaba | 47.8% | 2026-10-21 |
| 23 | GPT-6 Astra High | OpenAI | 46.7% | 2026-08-30 |
| 24 | Muse Wonder Max | Meta | 45.6% | 2026-10-14 |
| 25 | Gemini 3.7 Flash High | 41.1% | 2026-10-21 | |
| 26 | Grok 4.6 Extra High | xAI | 41.1% | 2026-10-21 |
| 27 | GPT-5.6 Terra Max | OpenAI | 41.1% | 2026-10-20 |
| 28 | Gemini 3.8 Flash High | 38.9% | 2026-09-02 | |
| 29 | GLM-5.2 Max | Zhipu AI | 36.7% | 2026-10-21 |
| 30 | DeepSeek V4 Pro 0813 Max | DeepSeek | 33.3% | 2026-10-20 |
| 31 | Claude Opus 4.6 Max | Anthropic | 33.3% | 2026-10-21 |
| 32 | Grok 4.5 High | xAI | 32.2% | 2026-10-21 |
| 33 | GPT-5.2 Extra High | OpenAI | 32.2% | 2026-10-21 |
| 34 | Claude Opus 4.7 High | Anthropic | 31.1% | 2026-08-12 |
| 35 | Mistral Large 4 Preview High | Mistral | 27.8% | 2026-10-21 |
| 36 | GPT-6 Luna Max | OpenAI | 26.7% | 2026-10-21 |
| 37 | Gemini 3.6 Flash High | 23.3% | 2026-10-21 | |
| 38 | GPT-5.5 Extra High | OpenAI | 23.3% | 2026-10-29 |
| 39 | GPT-5.4 Extra High | OpenAI | 23.3% | 2026-10-29 |
| 40 | GPT-5.6 Luna Max | OpenAI | 21.1% | 2026-10-21 |
| 41 | GPT-5.6 Sol High | OpenAI | 20.0% | 2026-08-10 |
| 42 | GPT-5.4 High | OpenAI | 15.6% | 2026-08-10 |
| 43 | GPT-5.5 High | OpenAI | 10.0% | 2026-08-10 |
| 44 | Gemini 3.1 Pro Preview High | 8.9% | 2026-08-10 | |
| 45 | GPT-5 High | OpenAI | 4.4% | 2026-10-21 |