AA SciCode: leaderboard
SciCode breaks 80 laboratory research problems from 16 scientific fields into subproblems, each needing a Python function that must pass scientist-written test cases. Artificial Analysis runs the 288 test set subproblems itself, leaving out the 50 validation ones, with the scientists' background notes included in the prompt. The number is the share of subproblems whose code passes, higher is better.
Metric: Accuracy (%). Source: artificialanalysis.ai. Saturation forecast: Around 2029. 235 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Muse Wonder Full (Max) | 62.62 | #10 (Muse Wonder Full) |
| 2 | Gemini 4 Argon (High) | 61.81 | #8 (Gemini 4 Argon) |
| 3 | MiMo-V2.6-Pro | 60.88 | #42 |
| 4 | Muse Wonder (Max) | 60.3 | #15 (Muse Wonder) |
| 5 | Gemini 3.7 Flash (Medium) | 59.84 | #23 (Gemini 3.7 Flash) |
| 6 | Muse Spark 1.3 (xHigh) | 59.72 | #24 (Muse Spark 1.3) |
| 7 | Kimi K3 (Max) | 59.49 | #30 (Kimi K3) |
| 8 | GLM-5.3 (Max) | 59.03 | #39 (GLM-5.3) |
| 9 | Step 5 Preview | 58.91 | #44 |
| 10 | Muse Spark 1.3 (Max) | 58.8 | #24 (Muse Spark 1.3) |
| 11 | Muse Spark 1.1 (xHigh) | 58.8 | #52 (Muse Spark 1.1) |
| 12 | Gemini 3.1 Pro (Preview) | 58.68 | #58 |
| 13 | Muse Spark Open (xHigh) | 58.45 | #54 (Muse Spark Open) |
| 14 | GPT-5.6 Sol (High) | 57.75 | #19 (GPT-5.6 Sol) |
| 15 | Grok 4.7 (High) | 57.75 | #25 (Grok 4.7) |
No result here: #3 Claude Opus 5.5, #9 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=aa-scicode · How It Works · Data refreshed daily, snapshot 2026-10-20.