Agentic Task Completion on Terminal-Bench Med. 2 (55 tasks)
88.2Pass@1AHE
Evaluation results
| Method | Links | |
|---|---|---|
| AHEBase Model=GPT-5.4 (high reasoning), Rollouts (k)=22026.04 | 88.2 | |
| CodexBase Model=GPT-5.4 (high reasoning), Rollouts (k)=22026.04 | 80 | |
| TF-GRPOBase Model=GPT-5.4 (high reasoning), Rollouts (k)=22026.04 | 79.4 | |
| NexAU0Base Model=GPT-5.4 (high reasoning), Rollouts (k)=22026.04 | 78.2 | |
| ACEBase Model=GPT-5.4 (high reasoning), Rollouts (k)=22026.04 | 78.2 | |
| terminus-2Base Model=GPT-5.4 (high reasoning), Rollouts (k)=22026.04 | 74.5 | |
| opencodeBase Model=GPT-5.4 (high reasoning), Rollouts (k)=22026.04 | 52.7 |