My limits

Two different things bound what you can run: credits (money, in your wallet) and quota (resources, set by a policy). This page is about the second.

The policy that applies to you

The card lists the effective ceilings, already resolved. Policies can be set per user, per group, per organization, and globally, and they resolve field by field: for each limit the most specific policy that sets it wins, and a blank falls through to the next scope. You see the result, not the four candidates.

Limit	What it caps
Concurrent sessions	Sessions running, paused, or queued at once
Queued sessions	How many may wait at the same time
VRAM / GPU cores	Totals across all your GPU sessions
CPU / Memory / Disk	Host resources across all your sessions
Storage	Total quota of all your volumes
Max runtime	After this, a session is terminated automatically
Idle timeout	After this much idleness, a GPU session is paused and its card returned
CPU session limits	The same caps applied separately to free CPU sessions

A blank entry means unlimited.

The same figures, as headroom rather than ceilings, appear on the dashboard and at the top of the session wizard — that is where you notice them, because a tier you cannot afford is marked there.

Requesting more

Request resource increase (on this page, the dashboard, and the wallet) opens a form with one field per ceiling. Fill in only what you need raised, each showing its current value, and give a reason.

The history below tracks what you asked for:

State	Meaning
Pending	With an administrator
Approved	The new ceiling is already in effect — this page shows the raised figure
Rejected	Declined, with the reason

Which error means which limit

Message	Cause	Fix
Quota exceeded	A policy ceiling — VRAM, CPU, storage	Request an increase here
Concurrency limit reached	Too many sessions at once	End or pause one, or request more
Insufficient credits	Money, not quota	Request credits
No GPU capacity	The cluster is full; nothing to do with your limits	Wait in the queue or take a smaller tier