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 | 0.4412 | $1.72 | 93.1M | 282k | 19h 4m | 430 | 2 |
| 2 | DeepSeek V4 Flash Vision Exp | 0.4247 | $2.96 | 173.2M | 352k | 17h 55m | 617 | 4 |
| 3 | DeepSeek V4 Flash Vision Exp | 0.4065 | $1.73 | 95.5M | 262k | 17h 34m | 448 | 5 |
| 4 | DeepSeek V4 Flash Vision Exp | 0.3803 | $1.85 | 99.9M | 306k | 18h 37m | 455 | 3 |
| 5 | DeepSeek V4 Flash Vision Exp | 0.2979 | $2.02 | 115.0M | 266k | 19h 32m | 508 | 4 |