Terminal Command Execution on Terminal-Bench 2.0
26.2AccuracyOpenThinkerAgent-32B
Evaluation Results
| Method | Links | |
|---|---|---|
| OpenThinkerAgent-32BTrain Size=100K, Method=SFT, Backbone=Qwen3-32B2026.06 | 26.2 | |
| Nemotron-Terminal-32BTrain Size=264K, Method=SFT, Backbone=Qwen3-32B2026.06 | 25.1 | |
| SA-SWE-32BTrain Size=4.5K, Method=RL, Backbone=Qwen3-32B2026.06 | 16.2 | |
| SWE-Lego-Qwen3-32BTrain Size=18K, Method=SFT, Backbone=Qwen3-32B2026.06 | 16.1 | |
| SERA-32BTrain Size=25K, Method=SFT, Backbone=Qwen3-32B2026.06 | 9.7 | |
| Qwen3-32BTrain Size=N/A, Method=N/A, Backbone=Qwen3-32B2026.06 | 7.5 | |
| DeepSWE-PreviewTrain Size=4.5K, Method=RL, Backbone=Qwen3-32B2026.06 | 4.9 |