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 | DeepSeek V4 Flash Vision Exp | 0.0268 | $2.15 | 112.2M | 397k | 7h 43m | 389 | 3 |
| 2 | DeepSeek V4 Flash Vision Exp | 0.0228 | $5.13 | 304.7M | 583k | 13h 30m | 712 | 7 |
| 3 | DeepSeek V4 Flash Vision Exp | 0.0094 | $6.17 | 373.9M | 627k | 11h 53m | 798 | 9 |
| 4 | DeepSeek V4 Flash Vision Exp | 0.0081 | $3.17 | 180.3M | 434k | 7h 52m | 554 | 2 |
| 5 | DeepSeek V4 Flash Vision Exp | 0.0056 | $3.93 | 222.7M | 553k | 14h 38m | 530 | 7 |