Not currently — once submitted, a job stays in the queue. A job that seems stuck for a long while usually isn't broken: when Leo detects a poor result on the backend, it automatically retries with varied parameters, and only gives up after several attempts. A long-running page therefore usually means the material is proving genuinely hard for the model — very unlike what it was trained on — rather than a fault in the system. If the job ultimately fails, your credit is refunded automatically.
