SciCode 1.0.1 (Artificial Analysis)
Code
Artificial Analysis' SciCode 1.0.1 evaluation over 288 test subproblems, with pass@1 averaged over three attempts. The current public data explicitly uses scicode_1_0_1; earlier unversioned SciCode observations are preserved separately.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Opus 5.5 | 66.9 |
| Claude Fable 5.1 | 63.1 |
| Claude Fable 5 | 61.0 |
| MiMo-V2.6-Pro | 60.9 |
| Gemini 3.7 Flash | 59.8 |
| Muse Spark 1.3 | 59.7 |
| Kimi K3 | 59.5 |
| GLM-5.3 | 59.0 |
| Step 5 Preview | 58.9 |
| Muse Spark 1.1 | 58.8 |
| Gemini 3.1 Pro Preview | 58.7 |
| GPT-5.6 Sol | 57.8 |
| Grok 4.7 | 57.8 |
| GPT-6 Sol | 57.6 |
| Muse Spark 1.2 | 57.4 |
| Gemini 3.8 Flash | 56.6 |
| GPT-6 Astra | 56.5 |
| Grok 4.6 | 56.5 |
| Claude Opus 5 | 56.4 |
| GPT-5.5 | 56.1 |
| GPT-5.6 Terra | 55.0 |
| Grok 4.5 | 55.0 |
| GPT-6 Luna | 54.6 |
| Opus 4.8 | 54.4 |
| Sonnet 5 | 54.3 |
| Qwen3.8 2.4T A95B | 54.1 |
| Gemini 3.5 Flash | 53.9 |
| GPT-5.6 Luna | 53.6 |
| Gemini 3.6 Flash | 53.4 |
| Qwen3.8-Max | 53.2 |
| GPT-5.5 Instant (2026-06-25 hosted snapshot) | 52.5 |
| GPT-5.4 Mini | 52.1 |
| Qwen3.8 Max (0902) | 52.1 |
| DeepSeek V4.1 Flash | 51.9 |
| Agnes 3.0 Flash (hosted) | 51.6 |
| GLM-5.3 Flash | 51.6 |
| Kimi K2.6 | 51.5 |
| GLM-5.2 | 51.2 |
| DeepSeek V4 Pro 0813 | 51.0 |
| DeepSeek-V4-Pro | 50.8 |