Cursor is changing how certain “frontier” models are billed for business customers, and it starts soon: beginning March 16th, Team and Enterprise accounts still on legacy request-based pricing will need to enable Max Mode to access specific high-end models.
The company says all other models remain unaffected, and that individual plans—as well as accounts already on the newer pricing introduced in June 2025—aren’t part of this shift.
What’s changing on March 16th
For Team and Enterprise customers on legacy per-request pricing, access to the following models will be gated behind Max Mode:
- GPT 5.3 Codex
- GPT 5.4
- Opus 4.5/4.6
- Sonnet 4.5/4.6
Cursor characterizes this as a move away from fixed-per-request billing for these models, tying their usage to Max Mode instead.
Why Cursor is tying frontier models to Max Mode
The reasoning is mostly about cost variance. Cursor notes that frontier models have become more capable, and that capability comes with longer runtimes, larger context windows, and significantly higher token usage per interaction. In that world, a “single complex request can vary widely in cost,” and fixed request-based pricing no longer tracks real usage.
So the stated goal is to transition these models to token-based billing, aligning costs with how much compute is actually consumed—matching the approach Cursor says it introduced for individual plans in the June 2025 pricing update.
Communication and what to expect for admins
Cursor says this was emailed to Team and Enterprise admins the prior week “for broader visibility,” and that Enterprise account owners will be contacted separately with account-specific details.
In the same thread, some users report that after a recent update, Max Mode appears to be forcibly turned on for several main models, prompting questions about whether this “effectively mean[s] a price increase,” along with requests for rollback options and alternative tools. Cursor also notes that new posts on the topic will be redirected or merged into the megathread, which will be updated with FAQs.
Original source: https://forum.cursor.com/t/megathread-frontier-models-moving-to-max-mode-for-legacy-team-enterprise-plans/155053
