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 | 0.1621 | $10.73 | 658.6M | 994k | 19h 13m | 1651 | 18 |
| 2 | DeepSeek V4 Flash Vision Exp | 0.0738 | $6.53 | 405.2M | 565k | 19h 26m | 994 | 10 |
| 3 | DeepSeek V4 Flash Vision Exp | 0.0483 | $6.64 | 411.0M | 525k | 16h 0m | 983 | 21 |
| 4 | DeepSeek V4 Flash Vision Exp | 0.0324 | $12.46 | 767.8M | 1.1M | 18h 4m | 1789 | 24 |
| 5 | DeepSeek V4 Flash Vision Exp | 0.0121 | $4.01 | 242.4M | 395k | 12h 19m | 777 | 16 |