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 | 0.9173 | $3.12 | 179.4M | 413k | 18h 25m | 558 | 6 |
| 2 | DeepSeek V4 Flash Vision Exp | 0.6906 | $1.80 | 93.9M | 332k | 18h 0m | 378 | 6 |
| 3 | DeepSeek V4 Flash Vision Exp | 0.4712 | $1.05 | 54.4M | 192k | 19h 29m | 342 | 2 |
| 4 | DeepSeek V4 Flash Vision Exp | 0.2438 | $1.59 | 84.5M | 269k | 19h 1m | 389 | 5 |
| 5 | DeepSeek V4 Flash Vision Exp | 0.0562 | $2.23 | 120.5M | 374k | 18h 41m | 436 | 3 |