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 | 0.9597 | $1.51 | 70.8M | 347k | 3h 23m | 283 | 4 |
| 2 | DeepSeek V4 Flash Vision Exp | 0.7308 | $2.17 | 110.7M | 423k | 3h 32m | 374 | 9 |
| 3 | DeepSeek V4 Flash Vision Exp | 0.0718 | $2.02 | 102.8M | 397k | 15h 29m | 346 | 4 |
| 4 | DeepSeek V4 Flash Vision Exp | 0.0700 | $1.50 | 70.3M | 352k | 6h 13m | 290 | 6 |
| 5 | DeepSeek V4 Flash Vision Exp | 0.0559 | $1.32 | 62.9M | 305k | 6h 32m | 278 | 4 |