Kiro Pricing: Free to $200 Credit Plans

Kiro is free for 50 credits a month. Paid plans are Pro at $20 (1,000 credits), Pro+ at $40 (2,000), Pro Max at $100 (5,000) and Power at $200 (10,000). Extra credits cost $0.04 each, and plan credits do not roll over. The first upgrade earns a $20 credit. Prices checked 2026-10-31 on kiro.dev.

Prices checked 2026-10-31

Source: kiro.dev/pricing. Prices exclude tax.

Divide the plan price by its credits and every paid tier comes to $0.02 a credit ($20 for 1,000, $40 for 2,000, $100 for 5,000, $200 for 10,000). Add-on credits cost double that, $0.04 each. In other words, the monthly allowance is half price compared with topping up. This is our arithmetic from the page.

For the tool itself, see Kiro. For other tools' prices, see AI coding agent pricing compared.

What a credit is

Kiro defines a credit as "a unit of work in response to user prompts." Three details from the page matter.

• Prompt size varies. A simple prompt can use less than one credit. A complex one, such as executing a spec task, usually costs more than one.
• Model choice changes the rate. Different models consume credits at different rates. Kiro's example: a task that uses X credits in Auto uses 1.3X credits through Sonnet 4.6.
• Metering is fine-grained. Credits are metered to two decimal places, so the least a task can consume is 0.01 credits.

Anything you ask Kiro to execute consumes credits, in vibe mode or spec mode, including spec refinement, task execution and agent hook execution. Workflows use the same credit model, so more complex workflows use more credits. Kiro Web uses credits at the same rate whether work runs in the IDE, CLI or web, with no separate charge for cloud compute. Kiro Web is available to Pro, Pro+, Pro Max and Power. Kiro powers carry no extra charge.

Usage shows in the subscription dashboard inside Kiro, updated at least every five minutes.

Auto and model choice

By default, prompts go to Auto, which uses a mix of frontier models with specialized models for tasks such as intent detection and caching, aiming at a balance of quality, latency and cost. You can pick a model yourself. The pricing page names OpenAI's GPT-5.6 (Sol, Terra and Luna), Anthropic's Claude models (Opus 5, Opus 4.8, Sonnet 5, Sonnet 4.6, Sonnet 4.5 and Haiku 4.5), and open weight models including DeepSeek 3.2, MiniMax M2.5, GLM-5 and Qwen3 Coder Next. Credit multipliers per model are on Kiro's models page.

For those models in detail, see Claude models, GPT models and DeepSeek models.

Buying more credits

How it works depends on how you signed in.

Monthly plan credits reset at the start of each billing month and do not roll over. Kiro processes paid subscriptions on the first day of each calendar month.

The sign-up bonus

The first time you upgrade to any paid plan with social login or AWS Builder ID (not AWS Identity Center), Kiro credits $20 toward your subscription. A valid credit card is required. The page gives a worked example: upgrade to Pro on 15 June and you get $20 x 15/30, which is $10, credited for the rest of June. On 1 July your card is charged the $20 July price minus that remaining credit, so $10.

Note the proration rule. The bonus is calculated on days left in the first month, and the first charge falls on the 1st of the next month. Upgrading on the 1st gets the largest first-month effect; upgrading on the 30th gets almost none of the bonus's value that month.

Teams, enterprise and regions

• Teams. Each developer needs their own subscription. Team plans have flexible overages at $0.04 per credit. Centralized billing, SSO, usage analytics and enterprise security controls are on the Enterprise tier.
• Enterprise. Purchased and billed through AWS rather than by billing address. Authentication runs through AWS IAM Identity Center or an external identity provider. The free tier is not available in Enterprise.
• GovCloud. Kiro pricing in AWS GovCloud (US) Regions is about 20 percent higher than commercial pricing, applies to every paid tier, and has no free tier.
• Taxes. Prices exclude VAT or sales tax. Japanese billing addresses attract Japanese Consumption Tax.
• Paid-plan regions. Social-login and Builder ID purchases need a supported billing country, and the page lists a long set of countries. Model availability also follows each model provider's country rules.
• Payments. All major credit cards.

Where you can use the subscription

Kiro lists these as covered: the Kiro IDE, Kiro CLI, Kiro Web, Kiro Crew, ACP-compatible IDEs, and automation in software development, such as reviews in CI/CD. It says plainly that using the subscription through third-party automation harnesses, naming OpenClaw as an example, that route requests outside Kiro's native interfaces, is not permitted. If you run an agent harness, read that rule before you pick Kiro; OpenClaw is a tool the page names.

Cost per credit, compared

Every tier costs the same per credit. That means you do not save per credit by moving up, you only avoid paying double for add-ons. The right size is the smallest plan whose allowance covers your normal month, because every credit beyond it costs twice as much.

A simple rule from that: if you regularly need more than 1,500 credits a month, Pro+ is cheaper than Pro with add-ons. Pro plus 500 add-on credits is $40 for 1,500 credits, the same money as Pro+ with 500 more credits to spare. Pro plus 1,000 add-on credits is $60 for 2,000 credits, against $40 on Pro+. Again, our arithmetic.

Worked monthly scenarios

Kiro does not publish how many credits a given task uses, so these scenarios are built on money, not tasks. They show what each plan costs for a month that uses a given number of credits, from the page's prices.

Unused plan credits are lost each month, so a plan that is too big wastes money. Using 800 credits on Pro Max would cost $100 for what Pro covers at $20. The reverse also bites: a plan that is too small pays $0.04 a credit on the overflow.

The sign-up bonus changes your first bill. A first upgrade to Pro on the 1st earns the full $20 credit proration for the days left, and the page's own example shows a mid-month upgrade ending with a $10 first charge on the 1st of the next month. After that the plan is its normal price.

What the page does not tell you

Four things we looked for and did not find on kiro.dev/pricing.

1. Credits per task. The page says a simple prompt can cost less than one credit and a spec task usually more than one, without numbers.
2. Model multipliers. Only one is stated: Sonnet 4.6 at 1.3 times the Auto rate in the example. The rest live on the models page.
3. Annual billing. The page describes monthly billing processed on the first of the month.
4. Refund terms. Not stated on the pricing page.

Treat the credit as a budget unit you learn by observation, not a price you can read off a rate card. Keep those in mind when comparing with tools that publish token-level prices, such as Claude Code pricing. With Kiro you are trusting the credit meter, and the only way to see how it behaves is to watch your dashboard.

Kiro against other tools

See Cursor pricing, GitHub Copilot pricing, Claude Code pricing and Codex pricing. Credits differ by vendor and are not interchangeable: a Kiro credit and a GitHub AI credit ($0.01) are different units.

Who should pick which

Our take

Kiro's page is among the plainest of the coding tools: a table of five rows, one credit price, one overage price. The thing to understand is that credits are a currency, not a count of prompts. A single spec task can burn more than one, a model switch can raise the rate by 30 percent, and the allowance does not roll over.

The decision rule is easy. Pick the plan whose included credits cover a normal month, and avoid add-ons as a habit, since each costs double. Use the dashboard for two weeks before committing to a larger plan. We have not tested Kiro, so we make no claim on quality or on how many credits a typical project consumes.

Checklist

• Check the dashboard weekly. Usage updates every five minutes.
• Prefer Auto unless you need a specific model.
• Do not buy add-on credits for a habit; step up a plan instead.
• Time the first upgrade for the $20 bonus.
• Do not route your subscription through third-party harnesses.
• Budget extra for GovCloud and for local tax.

Practical advice

Open the subscription dashboard in the first week and write down credits used per kind of task: a quick edit, a spec refinement and a spec task execution. After three or four tasks of each kind you have a personal rate card, which is more useful than any estimate. Then multiply by your expected weekly mix to choose a plan. Keep Auto as the default, since a task that costs X credits in Auto costs about 1.3X on Sonnet 4.6 by Kiro's own example, and use a named model only when the task needs it. If you are on a team, remember each developer needs a separate subscription, so budget per person and not per project.

One more way to think about it

Because every tier costs $0.02 a credit, choosing a Kiro plan is really choosing how large a prepaid monthly allowance to hold. A larger allowance does not buy a discount. It buys protection from the $0.04 overflow rate, and the price of that protection is waste whenever you under-use it. If your month-to-month usage is volatile, hold the smaller plan and pay overflow occasionally. If it is steady, size up to just above your normal month.