Glean’s research quota change moves the budget question

A usage quota is a blunt way to contain agent spending: it can stop both wasteful work and valuable work. Removing the quota shifts responsibility toward understanding which tasks consume resources and whether their results justify that consumption.

Glean’s October 13 release notes retire the per-user monthly Deep Research quota for organizations using the Glean Universal Model Key. Deep Research becomes a skill under standard Assistant terms; premium models on Enterprise Flex consume FlexCredits. Customer Key organizations retain the chat mode and its existing monthly quota.

No quota does not mean no cost

The same release makes Skills access a prerequisite for using the new skill. That means availability, permissions and billing are separate questions. An administrator should check the organization’s actual key and plan rather than assume every user received the same change.

The economic unit is a usable research result, not a started run. A report that needs extensive fact checking or repeats earlier work adds costs outside the model bill. A reasonable internal ledger would connect each task to its purpose, consumed credits, reviewer time and final disposition.

Replace the cap with a better measurement

Teams could begin with a small set of repeatable research tasks and compare accepted outputs with their existing process. Measure citations that support the claims, missing evidence and the time spent correcting results. The release does not disclose an end-to-end cost benchmark establishing a savings rate.

Less restrictive access could make research assistance easier to use where it is valuable. It could also make duplicate work less visible. The next evidence is whether users obtain more accepted decisions or artifacts per unit of total effort under the new terms, without shifting an unmeasured burden to reviewers.