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 | Grok 4.7 | 0.0456 | $436.54 | 434.7M | 834k | 19h 58m | 1732 | 29 |
| 2 | Grok 4.7 | 0.0000† | $129.68 | 133.6M | 655k | 14h 50m | 603 | 11 |
| 3 | Grok 4.7 | 0.0000† | $87.58 | 84.0M | 222k | 5h 40m | 303 | 8 |
| 4 | Grok 4.7 | 0.0000† | $51.73 | 43.4M | 216k | 3h 56m | 182 | 5 |
| 5 | Grok 4.7 | 0.0000† | $353.91 | 323.1M | 770k | 19h 55m | 1185 | 13 |
† score adjusted by manual review; the trace's Judge tab carries the measured score and the rationale.