TL;DR: Standard keeps its $40/month price but ships with substantially more included usage; a new Premium tier at $96/seat/month annual gives 5× that for 3× the price. Usage is now split into two separate pools — Cursor-native and third-party API — so billing surprises finally have a source. For most teams, assigning Premium to your 2–3 heaviest users and Standard to everyone else cuts the monthly bill nearly in half versus going all-Premium.

	Standard	Premium	Windsurf Teams
Best for	Regular coders, light-to-moderate agent use	Daily Claude Opus, overnight agent runs	Teams wanting flat-rate simplicity
Monthly price	$40/seat	$120/seat	$40/seat
Annual price	$32/seat	$96/seat	—
Usage vs. Standard	1× baseline	5×	Quota-based, not pooled
The catch	Third-party pool runs dry faster than Composer pool	$96/seat stings if usage is only moderate	No per-user seat type mixing

Honest take: On a 10-person team, put the 2 engineers who run agents all day on Premium and everyone else on Standard. Monthly cost drops to roughly $450 instead of $960 for all-Premium — and the other 8 developers won’t notice the difference.

---

What broke before and why this update exists

The August 2025 Teams pricing update moved agent usage from fixed request caps to variable token-rate billing. That change made economic sense — a 20-file refactoring run consumes far more compute than a single file rename, and flat request counts never reflected that. But the rollout came with one critical flaw: the usage meter was a single opaque number showing percentage-to-limit, with no breakdown of what was driving the charge.

Teams that defaulted to Claude Opus or GPT-4o for routine Composer sessions discovered their third-party API spend was 80% of their bill only after the invoice landed. Community forum threads filled up. The recurring complaint was not that Cursor was expensive — it was that Cursor was unpredictably expensive.

The June 2026 update has three practical goals: give Standard seats more included usage at the same price, create a Premium tier that actually serves power users without penalizing everyone else, and split the usage meter so admins can see the bill before it arrives.

The two usage pools, explained

Every Teams seat now carries two separate meters:

Pool 1 — Composer and Auto: Tracks consumption against Cursor’s first-party models and the Auto agent feature. This is where most developers spend the majority of their usage — code completions, Composer sessions, and agent tasks running on Cursor’s own inference stack. Cursor claims the Composer pool on a Premium seat covers a full month of heavy agent use for 99% of users.

Pool 2 — Third-Party API: Tracks usage of external models routed through Cursor: Claude Sonnet, Claude Opus, GPT-4o, Gemini Pro, and similar frontier models. This is where bills balloon for teams that set a frontier model as their default and never reconsider it.

The operational implication: a developer who uses Cursor’s native models heavily will almost never hit Pool 1. A developer who routes every Composer request through Claude Opus will drain Pool 2 independent of which seat type they have — the only difference is how large that bucket is before overages begin.

Neither seat type offers unlimited third-party API calls. Both pools have limits; Premium’s are 5× larger. If your entire team treats Claude Opus like autocomplete, the correct answer might be billing Anthropic directly via the API and setting a spending cap there, not buying Premium seats.

Standard at $32/seat (annual): who it actually serves

Standard costs $32/seat/month on an annual commitment or $40/seat/month month-to-month — identical monthly price to the pre-June plan, but with substantially more included usage in both pools.

What Standard covers:

• Full Cursor IDE with all features (Cursor Tab, Composer, Auto, BugBot)
• Significantly expanded Composer/Auto pool vs. pre-June baseline
• Third-party API pool for Claude and GPT usage (larger than before, but still finite)
• Centralized team billing, usage dashboard, spend alerts
• SSO (SAML/OIDC), RBAC, org-wide privacy mode

Standard handles the typical developer workflow: Cursor Tab completions running continuously, a few Composer sessions per day, occasional Auto agent runs for focused tasks, some Claude or GPT usage for complex reasoning. The pool expansion resolves the “hit the cap in week two” problem that plagued the August 2025 rollout for normal usage patterns.

Standard breaks down when a developer routes every session through a frontier model, runs Auto as an all-day background process, or regularly kicks off multi-file agent tasks that span hundreds of API calls. One or two developers on every mid-size engineering team fit this profile. The new Premium tier exists specifically for them.

Premium at $96/seat (annual): when the arithmetic holds

Premium delivers 5× the combined pool size of Standard at 3× the price. Per usage-unit, it’s cheaper than Standard-plus-overages for heavy consumers.

The break-even math: at the $64 annual-rate price gap between Premium ($96) and Standard ($32), overage charges need to exceed $64/month before Premium pencils out. At typical third-party API token rates, $64 in Claude Opus overages represents roughly 450K–700K output tokens beyond the Standard pool — achievable in a month of intensive agent development but not a baseline for most engineers. Developers who run overnight automated coding agents, who use Claude Opus for every Composer request across a full workday, or who regularly run cross-repository refactoring tasks are in this category.

Premium is not an “unlimited” plan. Both pools have ceilings, just 5× higher than Standard. Cursor’s own framing is that Premium covers “99% of heavy agent users” for a full month — which also implies the 1% who exceed it will still see overages.

What Premium is not the answer for: a developer who occasionally runs a long agent task but mostly uses Cursor Tab and standard Composer. That profile pays $64/month extra for headroom they rarely touch.

Mix-and-match: the structural change that actually matters

The June 2026 update lets Teams admins assign Standard or Premium seats per user. Previous Cursor Teams plans required the same tier for everyone. This is the change that makes the pricing math work for real teams.

A realistic breakdown for a 10-person team:

Seat count	Profile	Seat type	Annual rate
2	Senior engineers, daily agent runs, Claude Opus heavy use	Premium	$96/month
5	Mid-level engineers, daily Cursor Tab + regular Composer	Standard	$32/month
3	Junior / occasional Cursor use, mostly completions	Standard	$32/month

Monthly bill on annual plan: (2 × $96) + (8 × $32) = $448/month

All-Premium alternative: 10 × $96 = $960/month

All-Standard with overages for the two power users: $320 base + estimated $100–$200 in monthly third-party API overages = $420–$520/month — roughly equivalent to the mix-and-match approach but less predictable.

The mix-and-match approach delivers predictability without the overhead cost of blanket Premium. For a 20-person team with 4 heavy users, the savings scale proportionally: (4 × $96) + (16 × $32) = $896/month vs. $1,920 all-Premium.

The usage dashboard now shows per-user consumption broken down by pool, so admins can identify who is approaching limits and reassign seat types at the next billing cycle. Cursor also says it will surface recommendations when a user’s behavior consistently indicates they’d benefit from a seat upgrade — useful if accurate, a vendor upsell nudge if not.

The rebuilt spend-alerting system

The June update overhauled spend alerts. Admins can now configure dollar-based thresholds and receive notifications via Slack or email. The previous system showed a percentage bar that gave no signal about what the overage would cost in dollars.

Combined with the May 2026 admin controls that added soft limits at 50%, 80%, and 100% of usage with automated email triggers, the monitoring toolkit is now functional. It’s not as sophisticated as what a dedicated FinOps tool would provide — Cursor doesn’t expose per-model breakdowns at the API level — but it’s enough for a team lead to catch a runaway agent job before month-end.

For larger organizations with cloud cost governance requirements, this still may not satisfy procurement: there’s no webhook-based budget enforcement, no per-user spend caps enforced at the account level, and no real-time cost feed to pipe into cost management platforms. That gap likely explains why Enterprise (custom pricing, contact sales) exists as a separate tier.

Cursor Teams vs. GitHub Copilot Business vs. Windsurf Teams

The three main contenders at the team level are solving slightly different problems.

	Cursor Standard (annual)	Cursor Premium (annual)	Copilot Business	Windsurf Teams
Price/seat/month	$32	$96	$19	$40
Usage model	Two pools (Cursor + 3P API)	Two pools (5×)	AI Credits ($19/user)	Daily/weekly quotas
IDE requirement	Cursor IDE	Cursor IDE	All major IDEs	Windsurf IDE
Mix-and-match seats	Yes	Yes	No	No
Agent features	Auto, Composer	Auto, Composer	Copilot agent mode	Cascade agents
Admin spend controls	Dollar-based alerts, Slack/email	Same	Per-org policies	Usage dashboard only

Copilot Business at $19/user/month is the cheapest option and works inside VS Code, JetBrains, and most other editors — you don’t have to change the IDE. The tradeoff: Copilot is mid-June 2026 switching to AI Credits-based billing, which introduces its own usage-cap dynamics. See our GitHub Copilot June 2026 billing change breakdown for what that means for heavy teams. Copilot’s agent mode is improving but doesn’t match Cursor Auto’s depth of IDE integration for multi-file agentic tasks.

Windsurf Teams at $40/user/month uses daily and weekly quotas per user rather than monthly pools. There’s no shared pool that one power user can drain, which is a genuine advantage for team fairness, but there’s also no mechanism to give heavy users more headroom without moving everyone to Enterprise. If your team recently switched from Windsurf to Devin Desktop (Windsurf’s rebrand), check current pricing at windsurf.com directly — branding changes sometimes come with tier changes.

Cursor’s structural advantage is the pool split and mix-and-match flexibility. If your team has the bimodal usage distribution that most teams actually have — a few heavy users and a large majority of regular users — Cursor’s new model prices that distribution fairly. Flat per-seat competitors can’t match that for mixed teams.

When to skip Cursor Teams and use something else

Teams of 1–4 developers with no need for centralized billing, SSO, or admin controls should stay on individual Pro plans ($20/seat/month). The feature delta between Pro and Teams Standard is entirely administrative — the underlying model access and usage limits are comparable. At $32 vs. $20 per seat, you’re paying $12/seat/month for SSO, RBAC, org-wide privacy mode, and centralized invoicing. That’s worth it once finance needs a single invoice or security requires SSO; it’s not worth it before.

Teams running fully local AI coding stacks — Ollama, LM Studio, or similar — don’t pay per-token costs at all, making Teams pricing irrelevant. See our Continue.dev + Ollama setup guide, Cline + LM Studio walkthrough, and Aider + LM Studio guide for those workflows. For hardware guidance on running local models, runaihome.com’s VRAM tier guide covers what you need for different model sizes.

If your organization’s security team cannot approve cloud-based AI coding tools at all, Cursor Teams won’t pass that review regardless of pricing. In that case, a self-hosted stack with Cline or Continue.dev against a local or on-prem inference server is the only path.

The decision in one sentence

Buy Standard for everyone, upgrade to Premium only after 30 days of usage data show specific users consistently hitting the Standard pool limits — the dashboard now gives you exactly that signal.

---

FAQ

Is the new Standard seat the same monthly price as before?
Month-to-month, yes: $40/seat. The $32 rate is an annual-commitment price. The substantive change is that Standard now includes significantly more usage in both pools than the equivalent pre-June plan at the same price.

When do the new prices take effect?
Immediately for new Teams customers. For existing customers, changes apply at your first billing cycle starting July 1, 2026.

Can I switch individual users between Standard and Premium mid-month?
Per Cursor’s Teams docs, seat reassignments take effect at the next billing cycle. You won’t be charged for unused Premium days mid-month after a downgrade.

What happens when a user exhausts their pool?
Usage past the included limit is billed at on-demand token rates. The new spend alerting is designed to surface this before it lands on an invoice, but on-demand charges are still possible if alerts are not configured.

Does Cursor Teams include a trial period?
Cursor offers a 14-day trial for new Teams accounts. Individual developers can evaluate Cursor’s features on the free Hobby plan before committing.

Is there volume pricing for large teams?
The published pricing covers Standard and Premium seat types. Enterprise (custom pricing, contact Cursor sales) likely includes volume discounts for 50+ seat deployments, but terms are not public.

---

Sources

• Improvements to Teams Pricing — cursor.com/blog
• Team Pricing — Cursor Docs
• Updates to Teams and Auto (August 2025) — cursor.com/blog
• Plans for GitHub Copilot — GitHub Docs
• Windsurf Pricing 2026 — Verdent Guides
• Cursor Teams Upgrades Pricing for Predictability — StartupHub.ai
• Team admins now can control spend alerts and limits — Cursor Community Forum

Last updated June 6, 2026. Pricing and features change frequently; verify current state before purchasing.