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 | 0.4396 | $4.92 | 297.3M | 464k | 19h 11m | 681 | 11 |
| 2 | DeepSeek V4 Flash Vision Exp | 0.4384 | $3.53 | 204.1M | 419k | 17h 55m | 562 | 3 |
| 3 | DeepSeek V4 Flash Vision Exp | 0.3550 | $2.99 | 168.7M | 386k | 19h 32m | 453 | 3 |
| 4 | DeepSeek V4 Flash Vision Exp | 0.2720 | $3.61 | 208.7M | 436k | 14h 28m | 562 | 11 |
| 5 | DeepSeek V4 Flash Vision Exp | 0.2506 | $3.27 | 186.8M | 413k | 17h 5m | 493 | 5 |