My deployment is stuck at "requesting" — what does that mean?
Last updated: September 17, 2026
Your deployment is waiting for hardware. "Requesting" means Baseten is trying to provision an instance of the type you asked for, and one isn't immediately available.
The key idea: having access to a GPU type doesn't guarantee a machine is free at this moment. Self-serve deployments run on shared capacity pools, and popular GPU types can be momentarily exhausted.
What to try, in order:
Wait a few minutes. Capacity turns over constantly; many "stuck" deployments provision on their own.
Try a different instance type. If your model fits on more than one configuration, another type may have immediate availability.
Contact us with the deployment ID and instance type — we can check pool availability and tell you whether to wait, switch, or whether something else is wrong.