Scientific code generation benchmark as reported in the LLM Benchmark Matrix. Tests ability to write correct code for scientific computing problems across physics, chemistry, and biology.
This definition has not completed editorial and provenance review. Verify its metric, version, source, and evaluation configuration upstream before using it for a model-selection decision.
Tests domain-specific scientific programming — a gap in most coding benchmarks.
Matrix snapshot — we also have AA's own SciCode column. Very few models evaluated (3 scores).
Higher is better. Accuracy percentage (0–100%).
Matrix snapshot — we also have AA's own SciCode column.
| # | Model | Score |
|---|---|---|
| 1 | 58.9 | |
| 2 | 58.9 | |
| 3 | 56.1 | |
| 4 | 54.6 | |
| 5 | 54.6 | |
| 6 | 54.6 |