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 | Claude Fable 5.1 | 0.9662 | $443.40 | 798.0M | 1.8M | 18h 14m | 1802 | 27 |
| 2 | Claude Fable 5.1 | 0.9569 | $440.73 | 799.7M | 1.7M | 15h 10m | 1825 | 14 |
| 3 | Claude Fable 5.1 | 0.9473 | $365.73 | 654.8M | 1.6M | 15h 4m | 1521 | 20 |
| 4 | Claude Fable 5.1 | 0.9459 | $457.39 | 831.4M | 1.7M | 14h 37m | 1849 | 32 |
| 5 | Claude Fable 5.1 | 0.9053 | $549.23 | 1015.6M | 1.7M | 16h 2m | 2234 | 18 |
| 6 | GPT-5.6 | 0.7701 | $944.54 | 934.7M | 1.1M | 15h 1m | 2411 | 21 |
| 7 | GPT-5.6 | 0.7406 | $409.56 | 413.9M | 542k | 8h 34m | 1251 | 15 |
| 8 | GPT-5.6 | 0.7219 | $253.31 | 245.2M | 326k | 3h 19m | 679 | 6 |
| 9 | GLM-5.3 | 0.6108 | $356.80 | 1301.6M | 3.0M | 19h 1m | 2671 | 19 |
| 10 | GPT-5.6 | 0.6048 | $566.59 | 550.5M | 752k | 14h 19m | 1513 | 11 |
| 11 | Grok 4.6 | 0.5671 | $305.84 | 301.9M | 2.2M | 11h 0m | 1227 | 45 |
| 12 | Kimi K3 | 0.5098 | $230.68 | 643.2M | 1.9M | 20h 0m | 1545 | 3 |
| 13 | GPT-5.6 | 0.4981 | $490.17 | 481.7M | 531k | 5h 24m | 1274 | 32 |
| 14 | GLM-5.3 | 0.4715 | $281.50 | 1008.4M | 2.3M | 16h 6m | 2264 | 16 |
| 15 | GLM-5.3 | 0.4046 | $276.19 | 1010.3M | 1.8M | 15h 31m | 2070 | 21 |
| 16 | Kimi K3 | 0.3706 | $227.60 | 629.2M | 1.9M | 19h 47m | 1504 | 2 |
| 17 | GLM-5.3 | 0.3427 | $275.26 | 1018.0M | 1.7M | 17h 39m | 2090 | 39 |
| 18 | Grok 4.6 | 0.3420 | $868.26 | 872.5M | 3.4M | 16h 30m | 3114 | 373 |
| 19 | Kimi K3 | 0.3170 | $227.44 | 622.4M | 1.9M | 19h 52m | 1439 | 1 |
| 20 | Grok 4.6 | 0.2607 | $223.96 | 223.8M | 1.5M | 7h 20m | 882 | 18 |
| 21 | Grok 4.6 | 0.2416 | $205.23 | 195.4M | 1.9M | 7h 34m | 792 | 28 |
| 22 | Gemini 3.7 Flash | 0.2415 | $49.29 | 499.9M | 1.0M | 4h 48m | 1302 | 2 |
| 23 | Grok 4.6 | 0.2197 | $129.50 | 120.8M | 2.0M | 7h 52m | 540 | 19 |
| 24 | Gemini 3.7 Flash | 0.1895 | $40.63 | 436.9M | 982k | 3h 1m | 1134 | 2 |
| 25 | Gemini 3.7 Flash | 0.1724 | $88.83 | 943.5M | 2.1M | 6h 27m | 2449 | 1 |
| 26 | Gemini 3.7 Flash | 0.1698 | $29.14 | 318.3M | 646k | 2h 19m | 856 | 2 |
| 27 | GLM-5.3 | 0.1440 | $63.57 | 232.5M | 498k | 3h 14m | 487 | 6 |
| 28 | DeepSeek V4 Flash Vision Exp | 0.1348 | $19.48 | 1192.0M | 1.8M | 11h 51m | 2670 | 7 |
| 29 | Kimi K3 | 0.1336 | $196.38 | 508.8M | 1.8M | 19h 51m | 1209 | 4 |
| 30 | Gemini 3.7 Flash | 0.1224 | $55.17 | 592.3M | 1.3M | 5h 23m | 1507 | 5 |
| 31 | DeepSeek V4 Flash Vision Exp | 0.1138 | $28.55 | 1772.4M | 2.4M | 15h 1m | 3833 | 23 |
| 32 | Kimi K3 | 0.0909 | $184.74 | 495.2M | 1.5M | 19h 58m | 1154 | 2 |
| 33 | DeepSeek V4 Flash Vision Exp | 0.0534 | $18.46 | 1143.5M | 1.6M | 13h 42m | 2700 | 16 |
| 34 | DeepSeek V4 Flash Vision Exp | 0.0419 | $18.93 | 1164.8M | 1.7M | 11h 44m | 2764 | 29 |
| 35 | DeepSeek V4 Flash Vision Exp | 0.0319 | $17.22 | 1033.6M | 1.8M | 11h 35m | 2306 | 13 |
| 36 | Qwen3.8-Max | 0.0240 | $114.04 | 338.0M | 2.8M | 19h 40m | 884 | 6 |
| 37 | Qwen3.8-Max | 0.0224 | $119.80 | 401.3M | 2.9M | 20h 1m | 934 | — |
| 38 | Qwen3.8-Max | 0.0183 | $113.87 | 373.6M | 3.0M | 19h 42m | 990 | — |
| 39 | Muse Spark 1.2 | 0.0175 | $37.47 | 155.3M | 1.3M | 5h 29m | 460 | 7 |
| 40 | Qwen3.8-Max | 0.0159 | $129.76 | 393.1M | 2.8M | 19h 53m | 1102 | — |
| 41 | Qwen3.8-Max | 0.0002 | $139.67 | 464.7M | 2.7M | 19h 56m | 1203 | — |
| 42 | Inkling | 0.0000 | $3.82 | 3.9M | 21k | 15m | 103 | 2 |
| 43 | Inkling | 0.0000 | $6.62 | 6.9M | 28k | 22m | 168 | 1 |
| 44 | Inkling | 0.0000 | $8.48 | 8.8M | 35k | 24m | 171 | 2 |
| 45 | Muse Spark 1.2 | 0.0000 | $44.79 | 140.1M | 936k | 4h 21m | 543 | 4 |
| 46 | Inkling | 0.0000 | $10.64 | 11.1M | 32k | 19m | 193 | 1 |
| 47 | Muse Spark 1.2 | 0.0000† | $4.68 | 18.3M | 123k | 58m | 156 | 2 |
| 48 | Muse Spark 1.2 | 0.0000† | $53.19 | 95.9M | 776k | 4h 50m | 351 | 2 |
| 49 | Inkling | 0.0000 | $5.72 | 5.9M | 31k | 22m | 140 | 2 |
| 50 | Muse Spark 1.2 | 0.0000† | $10.59 | 43.3M | 223k | 2h 38m | 259 | 3 |
† score adjusted by manual review; the trace's Judge tab carries the measured score and the rationale.