Amazon Q Developer is on the way out, and that changes how you should read every number below. AWS announced the sunset in April 2026: new Free Tier accounts and new Pro subscriptions were blocked on May 15, 2026, Opus 4.6 was pulled from Pro on May 29, 2026, and the IDE plugins and paid subscriptions stop working on April 30, 2027. AWS points new users to Kiro. If you’re evaluating a coding agent today, you can’t sign up for this one; the pricing here is for teams already on it who need to plan a migration.
The parts of Q Developer that live inside AWS’s own surfaces are not affected — the Management Console, the AWS documentation site, the Console Mobile App, and the Slack and Microsoft Teams integrations all continue. What’s sunsetting is the IDE and CLI product that competes with Copilot and Cursor.
For existing accounts, the Free Tier is still a real monthly allowance rather than a trial: 50 agentic requests per month, where an “agentic request” covers any Q&A chat or agentic coding interaction through the IDE or CLI. That’s paired with Claude model access, general Q&A and basic error diagnosis inside the AWS Console, reference tracking, and the ability to suppress suggestions that closely match public code. It renews every month until end of support, but “until April 2027” is the ceiling now, not “indefinitely.”
Where Free Tier runs out fastest is on volume. Fifty agentic requests a month sounds workable until you’re running Q Developer as your primary coding assistant during active feature work, at which point you’ll hit the ceiling well before the month resets. That’s the entire pitch for Pro: $19 per user per month buys “increased limits” on agentic requests (AWS doesn’t publish an exact Pro-tier request cap, framing it instead as an expansion over Free), plus features Free doesn’t offer at all, like an admin dashboard for managing users and policy across an organization and IP indemnity, which matters if your legal team cares about liability from AI-generated code.
Q Developer also does something most coding agents don’t: automated Java and .NET version upgrades, where it analyzes a codebase and submits a transformation plan. This is billed separately from the general agentic request quota, in lines of code. Free Tier includes 1,000 LOC per month for transformation work. Pro includes 4,000 LOC per user per month, and importantly, that allowance pools at the AWS payer-account level rather than resetting individually per user. AWS’s own math: ten Pro subscriptions under one payer account get 40,000 pooled LOC a month before any overage kicks in. Exceed the pooled allowance and AWS charges $0.003 per line of code submitted beyond it, calculated and itemized on your bill alongside the bundled discount from your subscription count.
Before committing a team to Pro, understand how the billing actually triggers. The $19/user/month fee doesn’t activate the moment someone installs the toolkit or logs in. It triggers only when a user actually performs an activation action, like requesting an agentic coding plan, running a code transformation, or accepting a code completion. That means you can roll Q Developer out broadly and only pay for the users who actually engage with it, not everyone who has access. The flip side is unforgiving: cancel mid-month and AWS charges you for the full month regardless of when in the cycle you canceled, since access continues through the billing period as a safeguard against unauthorized use after cancellation. AWS’s own guidance is blunt about this — wait until the end of your billing cycle to cancel, don’t cancel mid-cycle expecting a prorated refund.
The practical question for teams still on it is no longer which tier to provision — it’s how much Pro spend is worth committing between now and April 2027. Java and .NET code transformation is the piece with a hard deadline: it stops working entirely at end of support, so any modernization work you were planning to run through Q Developer needs to finish inside that window or move elsewhere. Price the pooled transformation allowance against that timeline; a handful of large upgrade jobs can burn through 4,000 LOC per user fast, and the $0.003/LOC overage adds up on genuinely large codebases even though it looks trivial per line. For everything else, the migration to Kiro or another agent is the decision worth spending time on.