FrontierSWE
Frontier-difficulty software engineering tasks. Published by its authors, copied by Epoch AI.
| Category | Coding |
|---|---|
| Metric | Accuracy |
| Source | Epoch AI (evaluator) |
| License | CC BY 4.0 |
| Benchmark page | epoch.ai |
| Status | Scored |
| Weight | 1.2 within Coding |
| Fit | Discrimination 1.60, noise 0.28 |
Results
| # | Model | Provider | Result | Date |
|---|---|---|---|---|
| 1 | Claude Fable 5.2 Max | Anthropic | 69.9% | 2026-11-11 |
| 2 | GPT-6 Astra Max | OpenAI | 65.5% | 2026-11-11 |
| 3 | Claude Opus 5.5 Max | Anthropic | 62.3% | 2026-11-11 |
| 4 | Claude Sonnet 5.5 Max | Anthropic | 61.9% | 2026-11-11 |
| 5 | Gemini 4 Carbon High | 61.2% | 2026-11-11 | |
| 6 | Claude Fable 5.1 Max | Anthropic | 56.3% | 2026-11-11 |
| 7 | Gemini 4 Argon High | 55.0% | 2026-11-11 | |
| 8 | Claude Opus 5 Max | Anthropic | 52.0% | 2026-11-11 |
| 9 | Claude Fable 5 Max | Anthropic | 47.0% | 2026-11-11 |
| 10 | Claude Haiku 5.5 Max | Anthropic | 43.8% | 2026-11-11 |
| 11 | GPT-5.6 Sol Max | OpenAI | 32.2% | 2026-11-11 |
| 12 | GLM-5.3 Max | Zhipu AI | 30.2% | 2026-11-11 |
| 13 | Grok 4.7 Extra High | xAI | 29.5% | 2026-11-11 |
| 14 | Kimi K3 Max | Moonshot AI | 25.9% | 2026-11-11 |
| 15 | Muse Spark 1.3 Max | Meta | 25.8% | 2026-11-11 |
| 16 | Grok 4.6 Extra High | xAI | 25.3% | 2026-11-11 |
| 17 | Gemini 3.7 Flash High | 20.3% | 2026-11-11 | |
| 18 | Gemini 3.8 Flash High | 19.6% | 2026-11-11 | |
| 19 | GLM-5.3-Flash Max | Zhipu AI | 18.1% | 2026-11-11 |
| 20 | Qwen3.8 Max 0902 Extra High | Alibaba | 17.8% | 2026-11-11 |
| 21 | Qwen3.8 Max Extra High | Alibaba | 15.8% | 2026-11-11 |
| 22 | Muse Spark 1.2 Extra High | Meta | 12.0% | 2026-11-11 |
| 23 | Inkling Extra High | Thinking Machines | 4.1% | 2026-11-11 |