SciCode
Research-grade scientific coding problems.
- Published by
- SciCodedata via Epoch AI Benchmarking Hub
- Category
- Coding
- Index weight
- 0.5
- Models
- 158
- Data as of
- 9 Sept 2026
CC BY 4.0 — Epoch AI, ‘AI Benchmarking Hub’, epoch.ai/benchmarks. Mirrored boards credit their original publishers.
- 1Claude Fable 5.162.0%
- 2Claude Fable 560.2%
- 3Claude Fable 5.160.1%
- 4Gemini 3.1 Pro58.9%
- 5Kimi K358.7%
- 6Kimi K358.7%
- 7Muse Spark 1.158.2%
- 8Gemini 3.7 Flash57.9%
- 9Claude Fable 5.157.6%
- 10GPT-5.6 Sol56.9%
- 11Gemini 3.7 Flash56.8%
- 12GPT-5.456.6%
- 13GPT-6 Astra56.5%
- 14GPT-5.6 Sol56.5%
- 15GLM-5.356.5%
158 of 158
| # | |||||
|---|---|---|---|---|---|
| 1 |  Claude Fable 5.1Anthropic | 62.0% | – | 70.2 | – |
| 2 | Claude Fable 5Anthropic | 60.2% | – | 65.8 | – |
| 3 | Claude Fable 5.1Anthropic | 60.1% | – | 70.3 | – |
| 4 |  Gemini 3.1 ProGoogle | 58.9% | – | 63.9 | – |
| 5 |  Kimi K3Moonshot AIopen | 58.7% | Kimi K3 | 66.5 | – |
| 6 | Kimi K3Moonshot AIopen | 58.7% | – | 62.7 | – |
| 7 |  Muse Spark 1.1Meta | 58.2% | Muse Spark 1.1 (xhigh) | 64.9 | – |
| 8 | Gemini 3.7 FlashGoogle | 57.9% | – | 64.4 | – |
| 9 | Claude Fable 5.1Anthropic | 57.6% | – | 69.7 | – |
| 10 |  GPT-5.6 SolOpenAI | 56.9% | – | 68.0 | – |
| 11 | Gemini 3.7 FlashGoogle | 56.8% | – | 60.8 | – |
| 12 | GPT-5.4OpenAI | 56.6% | – | 61.3 | – |
| 13 | GPT-6 AstraOpenAI | 56.5% | – | 70.6 | – |
| 14 | GPT-5.6 SolOpenAI | 56.5% | – | 66.0 | – |
| 15 |  GLM-5.3Zhipu AIopen | 56.5% | – | 59.9 | – |
| 16 | Muse Spark 1.2Meta | 56.4% | – | 61.3 | – |
| 17 | GPT-5.6 SolOpenAI | 56.1% | – | 65.5 | – |
| 18 | GPT-5.5OpenAI | 56.1% | – | 63.7 | – |
| 19 | GPT-5.6 SolOpenAI | 56.0% | – | 67.5 | – |
| 20 | GPT-5.5OpenAI | 55.9% | – | 66.3 | – |
| 21 | GPT-6 AstraOpenAI | 55.7% | – | 68.6 | – |
| 22 | Claude Opus 5Anthropic | 55.7% | – | 66.8 | – |
| 23 | Claude Fable 5.1Anthropic | 55.7% | – | 65.9 | – |
| 24 | GPT-6 AstraOpenAI | 55.4% | – | 69.5 | – |
| 25 | GPT-5.6 SolOpenAI | 55.4% | – | 64.6 | – |
| 26 | Claude Fable 5.1Anthropic | 55.3% | – | 68.0 | – |
| 27 | Claude Opus 5Anthropic | 55.0% | – | 66.3 | – |
| 28 |  Grok 4.6xAI | 54.6% | – | 63.8 | – |
| 29 | Claude Opus 4.7Anthropic | 54.5% | – | 59.3 | – |
| 30 | Gemini 3.8 FlashGoogle | 54.4% | – | 63.5 | – |
| 31 | Claude Opus 5Anthropic | 54.3% | – | 69.3 | – |
| 32 | Gemini 3.8 FlashGoogle | 54.3% | – | 61.9 | – |
| 33 | GPT-6 AstraOpenAI | 54.2% | – | 67.4 | – |
| 34 | GPT-6 AstraOpenAI | 54.0% | – | 65.8 | – |
| 35 | Grok 4.5xAI | 54.0% | – | 54.2 | – |
| 36 | GPT-5.6 TerraOpenAI | 53.9% | – | 58.0 | – |
| 37 | Gemini 3.7 FlashGoogle | 53.6% | – | 62.5 | – |
| 38 | Gemini 3.8 FlashGoogle | 53.6% | – | 60.1 | – |
| 39 | Grok 4.6xAI | 53.6% | – | 58.7 | – |
| 40 | Claude Sonnet 5Anthropic | 53.6% | – | 50.7 | – |
| 41 | GPT-5.5OpenAI | 53.5% | – | 64.9 | – |
| 42 | GPT-6 AstraOpenAI | 53.5% | – | 64.4 | – |
| 43 | Kimi K2.6Moonshot AIopen | 53.5% | Kimi K2.6 | 60.6 | – |
| 44 | Claude Opus 4.8Anthropic | 53.5% | – | 59.4 | – |
| 45 | Gemini 3.5 FlashGoogle | 53.1% | – | 60.4 | – |
| 46 |  Qwen3 8Alibaba | 52.9% | Qwen3.8 Max | – | – |
| 47 | Gemini 3.6 FlashGoogle | 52.7% | – | 58.6 | – |
| 48 | GPT-5.6 LunaOpenAI | 52.5% | – | 56.5 | – |
| 49 | GPT-5.6 TerraOpenAI | 51.6% | – | 64.2 | – |
| 50 | Grok 4.6xAI | 51.6% | – | 62.3 | – |
| 51 | GPT-5.5OpenAI | 51.6% | – | 62.3 | – |
| 52 | Muse SparkMeta | 51.5% | – | 65.4 | – |
| 53 | Kimi K3Moonshot AIopen | 51.2% | – | 58.8 | – |
| 54 | Claude Opus 5Anthropic | 50.7% | – | 63.1 | – |
| 55 | GPT-5.6 LunaOpenAI | 50.7% | – | 58.1 | – |
| 56 | GLM-5.2Zhipu AIopen | 50.5% | – | 57.6 | – |
| 57 | MiMo-V2.5-ProXiaomiopen | 50.2% | – | 59.0 | – |
| 58 | Grok Build 0.1xAI | 50.2% | Grok Build 0.1 0616 | 56.1 | – |
| 59 | GPT-5.6 TerraOpenAI | 50.1% | – | 63.1 | – |
| 60 | GPT-5.6 LunaOpenAI | 50.0% | – | 59.3 | – |