Since June 10 I’ve had fine-tuning jobs on gpt-4.1-mini-2025-04-14 sit in queued for ~19–24h and never transition to running — no training, no trained_tokens, estimated_finish stays null.
Timeline (UTC):
- Job A — submitted Jun 10 ~18:48, sat queued ~19h with no progress, cancelled.
- Job B (resubmit) — submitted Jun 11 13:41, still queued ~24h as of writing.
For both: files validated successfully within seconds (Files validated, moving job to queued state), no error, and it’s the only fine-tune job in the org. Cancelling + resubmitting did not help (re-enters the same queue).
This is the base model that, per other forum threads, usually starts almost immediately — so ~24h queued seems abnormal. Is there a current capacity issue or incident affecting gpt-4.1-mini fine-tuning? Is there anything actionable on my side, or is this purely queue/capacity?
Happy to share the ftjob-… IDs and org ID with staff via DM.