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 4.6 | 0.1206 | $114.81 | 93.3M | 882k | 19h 41m | 426 | 10 |
| 2 | Grok 4.6 | 0.0335 | $207.50 | 166.7M | 1.1M | 18h 40m | 692 | 23 |
| 3 | Grok 4.6 | 0.0122 | $159.04 | 128.5M | 975k | 18h 46m | 569 | 20 |
| 4 | Grok 4.6 | 0.0043 | $218.28 | 183.9M | 1.3M | 14h 37m | 752 | 15 |
| 5 | Grok 4.6 | 0.0000† | $57.50 | 52.5M | 468k | 7h 59m | 252 | 3 |
† score adjusted by manual review; the trace's Judge tab carries the measured score and the rationale.