JetBrains AI Pricing Change Flips the Cursor Comparison

JetBrains AI Assistant now runs four tiers on a credits system instead of two flat unlocks — and the free tier costs $0, not $10.

The change matters because the most-cited reason to pick Cursor over JetBrains AI was price parity: same $20 a month, but Cursor includes the editor while JetBrains AI was an add-on to an IDE subscription you already paid for. DevTools Review, which ran hands-on testing across 11 AI coding tools, has publicly retracted that part of its own verdict.

The new tier structure

As reported in the retraction, the lineup is now:

Tier	Individual	Organization	Credits
AI Free	$0	—	3 AI Credits per 30 days, plus unlimited local completion via Ollama or LM Studio
AI Pro	$10/mo	$20/mo	10 credits (individual), 20 credits (org)
AI Ultimate	$30/mo	$60/mo	35 credits (individual), 70 credits (org)
AI Enterprise	Custom		—

The old description the verdict was built on — Personal at $10/month, Commercial at $20/month, no free option — no longer describes what is on offer.

The detail that actually flips the comparison

Not the tier count. It is the bundling: multiple independent trackers now report that AI Pro ships free with the All Products Pack and dotUltimate subscriptions. That is precisely the "add-on to a subscription you're already paying for" cost the original comparison charged against JetBrains.

For a developer already on All Products Pack or dotUltimate, the effective added cost of JetBrains AI is $0, not $20. The pricing round does not narrow — it inverts. Cursor at $20 including the editor stops being the cheaper option against an assistant that costs a team nothing extra on a license it already renews.

How firm is this?

Firm enough to act on, not firm enough to quote as vendor policy. DevTools Review is explicit that jetbrains.com was unreachable from its verification tooling at the time of writing, so it could not quote the vendor pricing page directly. What it has is consistent corroboration across four independent trackers — aiproductivity.ai, stackpick.net, uragent.org and checkthat.ai — plus a JetBrains community-forum thread describing the bundling firsthand.

The reviewers' own framing: treat the numbers as very likely correct, not vendor-verified. If you are about to move a team's seats on this, confirm the bundle on your own billing page first — that is a two-minute check that resolves the entire question for your specific plan.

What the change does not touch

The capability findings stand. Cursor's Tab completion is still described as faster and more aggressive. Junie, JetBrains' agent, still wins on inspection-clean correctness for Java and Kotlin because it executes through IntelliJ's actual refactoring engine rather than making text-level edits.

The concrete test behind that claim: both agents were asked to add pagination support to a set of Spring Boot REST endpoints. Junie found the endpoints, built a shared utility, updated the service and repository layers, and fixed two test failures on its own. That mechanism advantage is independent of what either tool costs.

What to watch

A credits system moves the buying question from "what does a seat cost" to "how many credits does our actual usage burn" — and 10 credits a month on AI Pro is a number that only means something once you have measured a week of real agent runs against it. Watch for JetBrains publishing per-model credit consumption, and re-check the All Products Pack bundling on the vendor's own page before it becomes the basis of a renewal decision.