Get quotas
Get the organization’s quotas. Each key of quotas holds its limit and current usage. max_gpus_per_instance, the largest GPU count one instance may have, caps each instance on its own, so it has a limit and no usage.
Usage is counted when the request is made. The GPU quotas count GPUs on running instances and running sandboxes together; a queued, building or starting resource already counts. instances counts every instance that is not deleted, running or stopped. The sandbox quotas (sandboxes, sandbox_memory_gib) are always present in quotas, 0 when the organization has none. The organization has Sandbox access exactly when quotas.sandboxes.limit is above 0.
standard_gpu_types lists the GPU types that count toward standard_gpus; every other GPU type counts toward premium_gpus. With usage=false the response holds limits only: no usage is counted and no quota has usage.
tier is the organization’s quota tier, which decides every limit the organization has no override for. overrides lists the limit keys the organization has its own override for, in a fixed order, and is an empty array when it has none. A first successful payment moves an organization on the student or new tier to a paid tier, which raises each limit the paid tier sets higher, except the keys in overrides; any other limit is raised only by support.
Authorizations
Response
OK
Overrides lists the limit keys the organization has its own override for, in a fixed order; empty, never null, when it has none. Those limits do not follow the tier, so only support raises them. It is an open list of strings, not an enum: a later version may add a key, which callers should ignore if they do not know it.