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 | 0.6244 | $2.53 | 133.2M | 457k | 10h 50m | 402 | 7 |
| 2 | DeepSeek V4 Flash Vision Exp | 0.5485 | $2.27 | 120.4M | 400k | 9h 24m | 405 | 3 |
| 3 | DeepSeek V4 Flash Vision Exp | 0.4883 | $3.26 | 177.3M | 519k | 9h 8m | 426 | 7 |
| 4 | DeepSeek V4 Flash Vision Exp | 0.4768 | $2.33 | 127.1M | 362k | 12h 24m | 407 | 4 |
| 5 | DeepSeek V4 Flash Vision Exp | 0.4348 | $2.35 | 127.0M | 382k | 9h 0m | 395 | 7 |