Every run: the agent's full trajectory, its submissions over the 20-hour budget, and the final score.
| # | Model | Reward | Cost | In tok | Out tok | Time | Steps | Submits |
|---|---|---|---|---|---|---|---|---|
| 1 | GPT-6 Astra | 0.9601 | $2439.09 | 1157.2M | 2.2M | 16h 56m | 2671 | 30 |
| 2 | GPT-6 Astra | 0.9508 | $2034.34 | 972.8M | 1.7M | 15h 13m | 2216 | 42 |
| 3 | GPT-6 Astra | 0.9472 | $895.41 | 422.2M | 827k | 6h 22m | 958 | 14 |
| 4 | GPT-6 Astra | 0.9426 | $2766.39 | 1322.7M | 2.4M | 19h 53m | 2984 | 74 |
| 5 | GPT-6 Astra | 0.8965 | $171.46 | 81.1M | 258k | 2h 1m | 277 | 7 |