↓ Skip to main content
  1. Articles/

Cursor Splits Its Teams Pricing in Two — and Bets $120/Month That You're the Problem

·891 words·5 mins·
Florent Clairambault
Author
Florent Clairambault
CTO & software engineer — writing daily about spec-driven development and agentic coding

Cursor Splits Its Teams Pricing in Two — and Bets $120/Month That You’re the Problem
Cursor rolled out a new Teams pricing structure in late June, and it starts landing on existing customers’ bills this week. Every renewal cycle beginning July 1, 2026 pulls existing Teams accounts into the new system automatically — no opt-in, no warning email required beyond whatever Cursor sends at renewal. If you manage a Cursor Teams account, this is the week it changes under you.

What Actually Changed
#

The old Teams seat bundled all usage — first-party models, third-party models, everything — into a single pool that ran out unpredictably and triggered overage charges when it did. The new structure splits every seat into two separate pools:

  • Composer and Auto — a generous allocation reserved for Cursor’s own first-party models and its Auto routing mode
  • Third-Party API — a separate, smaller allocation for calls to external models (Claude, GPT, Gemini, etc.)

On top of the split, Cursor introduced a second seat tier. Standard stays at $32/month annual ($40 monthly) with, per Cursor, “significantly more usage, with no change in cost” versus the old single-pool seat. Premium is new: $96/month annual ($120 monthly) — 3x the price — for 5x the Composer/Auto usage. Cursor says the Premium pool should cover a full month of heavy agent usage for 99% of users, and the admin dashboard now ships real-time usage visibility with Slack/email threshold alerts so a team doesn’t find out it blew through its pool at the invoice.

The Rationale, in Cursor’s Own Words
#

Cursor’s stated justification traces back to its own Developer Habits Report: a small subset of heavy users generates the majority of a team’s spend and cost volatility. The two-tier seat structure is explicitly designed to let those power users self-select into a plan built for their usage pattern — pay 3x, get 5x — rather than have their consumption blow up the shared pool (and the bill) for everyone else on Standard.

Cursor also frames this as a net win: the company estimates the changes lower costs for about 90% of teams. That’s plausible on its face — most seats probably aren’t heavy agent users, and giving Standard seats more headroom at the same price is a real improvement for that majority. But “90% pay less or the same” is also the framing you’d choose if you were introducing a new top tier priced to capture the 10% who were already your least profitable, most support-intensive customers. Both things can be true at once.

Why This Keeps Happening to Cursor Specifically
#

This is not Cursor’s first pricing restructuring in the past year, and it won’t obviously be the last. The pattern is consistent: usage-based agentic coding is expensive to run, actual usage is heavy-tailed (a handful of power users eat a disproportionate share of any shared pool), and IDE-centric tools built around interactive, human-supervised sessions have less predictable per-user cost than tools built for scoped, bounded agentic tasks. Every time Cursor tries to solve unpredictable costs with a new pricing tier, it’s implicitly admitting that the previous tier’s usage model wasn’t priced correctly for how people actually use the product.

Compare that to how Claude Code’s pricing moves. When Claude Sonnet 5 shipped as the new default model on June 30, the change was a straight token-rate adjustment — $2/$10 per million tokens through August 31, then $3/$15 — applied uniformly, with no new seat tier, no usage-pool split, and no admin dashboard rework required to understand what changed. That’s not because Anthropic doesn’t also have heavy users who cost more to serve; it’s because Claude Code’s pricing model was built around metered token consumption from the start, so absorbing a new model generation is a rate change, not a structural one. Cursor’s IDE-first, seat-based heritage means every attempt to control agentic cost has to be retrofitted onto a pricing model designed for a different product era — human-driven autocomplete, not autonomous multi-step agents burning tokens in the background.

What This Means If You Run a Cursor Teams Account
#

Check your renewal date. If it falls on or after July 1, 2026, you’re moving to the new structure automatically, and it’s worth auditing your team’s actual usage distribution before that happens rather than after the first surprising invoice. Cursor’s own admin dashboard now gives you the data to do that — usage visibility per pool, per seat, with alerting — so there’s no excuse for finding out the hard way which of your engineers needs a Premium seat and which are fine on Standard.

The more durable takeaway is the same one this blog keeps returning to: usage-based AI tooling costs are not a solved problem anywhere in the industry, but the tools built agent-native from day one are absorbing that volatility with rate changes, while the tools retrofitting agentic workflows onto older subscription models keep needing new tiers, new pools, and new dashboards to manage it. That’s not a moral failing on Cursor’s part — it’s an architectural one, and it’s the same one that shows up every time Cursor ships a fix for a pricing problem Claude Code’s metering model never had in the first place.


Sources:

AI Tools

Anthropic's Billing Split Is Live: The 175-to-1 Subsidy, Sam Altman's Counterpunch, and the Community Notes Correction

The Agent SDK credit split went live today. Community analyses reveal Max 20x subscribers were receiving up to 175x more compute than they paid for. On the exact same day Anthropic announced the change in May, Sam Altman offered enterprise customers two months of free Codex usage to switch. Here is what the Day 1 numbers actually look like.

Jun 15, 2026 · 7 min read

AI Tools

Anthropic's June 15 Billing Split: When Your CI Pipeline Becomes a Line Item

Starting June 15, Anthropic moves all programmatic Claude usage — Agent SDK, claude -p, Claude Code GitHub Actions, and third-party apps — to a separate monthly credit pool. Pro users get $20; Max 20x users get $200. The flat-rate agentic AI era is ending on both sides of the market.

Jun 6, 2026 · 6 min read

AI Tools

OpenAI Just Built an IT Services Company. That's an Admission.

OpenAI launched a $4B+ PE-backed deployment company on May 11, acquiring AI consultancy Tomoro and embedding 150 engineers into enterprise clients. The structure tells a story: if models alone were enough to win enterprise, you wouldn’t need a 1,000-person professional services arm.

May 13, 2026 · 6 min read

AI Tools

Cursor 3: Agent-First Branding, IDE-Last Architecture

Cursor 3 shipped a genuinely redesigned interface built around parallel agents. The Agents Window, Design Mode, /worktree, and /best-of-n are real features with real uses. But ‘agent-first’ describes the UI layer, not the architecture — and the distinction matters more than Cursor’s marketing suggests.

Apr 10, 2026 · 6 min read

AI Tools

Claude Sonnet 5: Anthropic's 'Most Agentic Sonnet Yet' Closes the Gap With Opus — For Half the Price

Anthropic shipped Claude Sonnet 5 on June 30 as the new default model in Claude Code and across all plans — native 1M-token context, 63.2% on agentic coding evals (vs. 58.1% for Sonnet 4.6), and introductory pricing of $2/$10 per million tokens through August 31.

Jul 1, 2026 · 5 min read