Glean adds Auto Routing to pick a model for each request

Glean launched Auto Routing, which matches each request to a model based on what the task needs, balancing speed, answer quality and cost.

Glean says about a quarter of requests are simple lookups, like finding a policy or a doc, and those go to a fast, lightweight path. The rest go to the main agent, with stronger models brought in only for work where Glean's testing showed a difference, such as interactive dashboards or exhaustive analysis.

Admins set a policy for the whole company or per department. Economy keeps everything on the most efficient models, Balanced is the recommended default, and Frontier uses the most capable models throughout. If an admin turns a model off, routing uses the next available one, and budget guardrails can step usage down as a team nears its monthly limit.

Glean says the routing is backed by an evaluation of 5,000 real enterprise tasks replayed across 37 model and reasoning configurations from 11 models.