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 | 0.6122 | $2035.40 | 958.4M | 1.9M | 17h 45m | 2132 | 53 |
| 2 | GPT-6 Astra | 0.6063 | $2124.53 | 1006.7M | 2.1M | 18h 49m | 2308 | 42 |
| 3 | GPT-6 Astra | 0.5422 | $1208.04 | 578.6M | 1.3M | 19h 17m | 1389 | 24 |
| 4 | GPT-6 Astra | 0.4322 | $274.97 | 132.6M | 360k | 5h 47m | 407 | 7 |
| 5 | GPT-6 Astra | 0.4121 | $327.78 | 158.4M | 411k | 6h 16m | 482 | 10 |