Agentic Coding on SWE-Bench Pro (Score, Relative Change)
32.7ScoreSAR
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SARAgent=CodeAct, Projection=top-1% spectral-rank2026.07 | 32.7 | 10.21 | |
| Baseline (In-house)Agent=CodeAct2026.07 | 29.7 | — |
Loading the SOTA2 catalog…
| Method | Links | ||
|---|---|---|---|
| SARAgent=CodeAct, Projection=top-1% spectral-rank2026.07 | 32.7 | 10.21 | |
| Baseline (In-house)Agent=CodeAct2026.07 | 29.7 | — |