All-task winner: none yet. Current adapter-smoke winner: poolside/laguna-xs-2.1.
| Family | Tasks | Prod passed | Runnable now | Blocked | Coverage |
|---|---|---|---|---|---|
spreadsheetbench-v1-full-912 | 912 | 0 | 912 | 0 | |
spreadsheetbench-v2-full-321 | 321 | 0 | 321 | 0 | |
bankertoolbench-full-100 | 100 | 0 | 100 | 0 | |
accounting-live-proofloop | 4 | 0 | 4 | 0 | |
notion-live-proofloop | 4 | 0 | 4 | 0 | |
proximitty-underwriting-pr0 | 4 | 0 | 4 | 0 | |
finch-prod-proxy-task | 1 | 1 | 1 | 0 | |
finauditing-prod-proxy-task | 1 | 1 | 1 | 0 | |
workstreambench-prod-proxy-task | 1 | 1 | 1 | 0 | |
noderoom-multi-user-conflict | 6 | 1 | 6 | 0 |
| Model | Adapter smoke | Est. cost | Measured cost |
|---|---|---|---|
z-ai/glm-5.2 | 3/3 | $0.2454 | $0.2450 |
deepseek/deepseek-v4-flash | 1/3 | $0.0683 | $0.0690 |
poolside/laguna-xs-2.1 | 3/3 | $0.0326 | $0.0330 |
qwen/qwen3.7-plus | 3/3 | $0.1994 | $0.1990 |