JetBrains still prices AI access apart from the IDE

The setup

JetBrains built its reputation on static analysis inside IDEs like IntelliJ IDEA, sold under a license separate from any AI feature. Its current AI product page, as retrieved 16 September 2026, describes JetBrains AI as a layer on top of those IDEs: a subscription giving unified access to leading AI models, alongside an explicit alternative of connecting a person's own provider keys, an approach it calls bring your own key. A companion company blog post states the underlying JetBrains AI Platform, an internal router, has powered JetBrains AI for customers and individual developers since 2023, dating the add-on three years before this description.

What the documents show

The AI product page frames the offering as an ecosystem rather than a single toggle: in-IDE features, a standalone coding agent called Junie, an agentic environment called Air, and a governance layer called JetBrains Central for organizations running several AI tools at once. On training data, the page states that if a customer chooses to share data, it is used exclusively to improve JetBrains tools and train the company's own models, such as Mellum, and is never shared with third parties. That is JetBrains describing its own practice, not an audited claim, paired with an assurance that enterprise deployments can keep data fully local, with JetBrains never accessing it in those setups.

The friction

The August 2026 post is unusually candid about cost on the same product line: JetBrains' own AI development expenses rose roughly tenfold over six months, attributed to wider adoption of terminal-based coding agents alongside the in-IDE assistant and a jump in usage once newer models arrived. The post describes the resulting administrative load in plain terms, several people needing to sign off on each request and tickets piling up, before JetBrains built its own internal usage-tracking tool. Neither page states a specific price for an individual seat, limiting how precisely a team can budget from the pages alone.

What changed in the work

What the sources support is that the add-on decision from 2023 did not stay simple. A team that adopted JetBrains AI as one line item now faces a wider menu, bring-your-own-key access, a subscription, a separate coding agent, and a separate agentic environment, each with its own cost profile, and JetBrains' own account of managing that sprawl internally is evidence the complexity is real rather than a documentation artifact. It is an editorial read to say the setup cost moved from choosing a plan to choosing among several products, but that shift is what the two pages together describe.

• Which of JetBrains' AI surfaces, the in-IDE assistant, Junie, or Air, is this team actually paying for, and separately from what?
• Is usage running through JetBrains' own subscription, a bring-your-own-key account, or both, and is anyone tracking the total?
• Does the no-third-party-sharing claim on the AI page match what a specific enterprise contract actually states?

A vendor admitting its own AI bill grew tenfold in six months is a useful corrective to any assumption that the AI add-on is still a single, predictable cost.