Every real run, newest first unless a column says otherwise. Open one to see the page the agent wrote, its violations and what each step cost.
Only runs of rule raw-values/raw-color (remove).
| Task | Model | Context | Design system | Quality | Done | Steps | Cost | When ↓ |
|---|---|---|---|---|---|---|---|---|
| loan-contract | claude-haiku-5.5 | none | 35 | 9 | Done | 3 | $0.0028 | Nov 8, 23:51 |
105 runs match.
| kyc-form | claude-haiku-5.5 | none | 31 | 5 | Done | 4 | $0.0034 | Nov 8, 23:51 |
| statement | gpt-6-luna | none | 115 | 17 | Done | 5 | $0.0049 | Nov 8, 23:10 |
| statement | gpt-6-luna | none | 146 | 25 | Done | 4 | $0.0026 | Nov 8, 22:29 |
| statement | nemotron-3.5-lightning-free | none | 20 | 20 | Incomplete | 3 | $0.0000 | Nov 8, 22:06 |