Cursor Router is available today across all surfaces on Teams and Enterprise pla

Cursor's router chooses a model for each request using the task, surrounding context and estimated complexity. Its launch account describes sending routine work to cheaper models while retaining more capable options for demanding work.

For three large early-access accounts, Cursor reports spending 30–50% less than the same traffic would have cost entirely at Anthropic's Opus 4.8 API rates. Quality assessment used signals including user corrections and how much generated code remained. This is a vendor-reported comparison against a priced alternative, with behavioral measures of usefulness rather than a proof of code correctness. Administrators can also restrict which models the router uses.

“In online A/B tests across millions of requests, Cursor Router delivered frontier-quality performance at 60% savings.”

Cursor

“Roughly 60% of developers using Cursor pick a single model as their daily driver. This results in routine work being completed at frontier prices, and AI spend growing much faster than output quality.”

Cursor

“We trained Cursor Router on 600k+ live requests and evaluated performance in an online A/B test across millions of live requests directed by Cursor Router, optimizing for user satisfaction (AFC) as a reward.”

Cursor

“In early access, three high-volume accounts with thousands of users saved 30%–50% on Auto-routed requests versus routing everything to Opus 4.8, with no decrease in quality.”

Cursor

“For a single commit, we observed Cursor Router had a lower cost per commit of $6.76 for Intelligence mode and $4.63 for Balance.”

Cursor