Amazon Quick pricing

An agent hour is a single unit that measures the time used by all AI features in Quick, including Chat, Research, Flows, Automate, artifact generation (such as PowerPoint, Word, and Excel), and custom AI-powered agents or apps created by users. Agent hours are consumed regardless of the interface you use: the desktop app, web, mobile app, or extensions. Usage is metered per second.

All agentic AI activity you initiate counts toward agent hours. That includes chatting with Quick’s out-of-the-box agent as well as curating your own custom agents on top of the default agent and using them either in real time or for scheduled tasks. Whichever you use, the time Quick spends working on your request is metered per second and counts against your agent hours.

Agent hours are metered per second and sent to the AWS metering system in near real time. Administrators who monitor usage on Enterprise accounts through AWS Cost Explorer may see a delay of up to 8 hours before usage is reflected. Users can see their own usage in the Quick interface under Mission Control.

Not necessarily. Agent hour usage depends on the complexity of the task and the mode selected (e.g., Fast, Balanced, Smart). Agent hours are metered per second, and only while Quick is actively working. Idle time between your interactions does not count. For a simple chat where you ask questions and review the results, your usage is likely at or below the length of your session. For a complex task that works across multiple tools and runs steps in parallel, 15 minutes of wall-clock time can use more than 15 minutes of agent hours. For example, a daily morning status report set to generate at 8 AM draws down your hours on its own each time it runs. Usage varies by person, but it becomes predictable once your patterns settle, and you can track consumption across all your interfaces in the runtime view under Mission Control.

There is no difference in terms of pricing or how usage is counted. Desktop, mobile, web, and extensions are all interfaces to the same agents.

Quick lets you choose how much reasoning power to put behind a task. Fast is the lightest option and consumes agent hours at the lowest rate, Balanced sits in the middle, and Smart applies the most reasoning and consumes at the fastest rate. You pick the mode that fits the task, so a quick lookup can stay light while a complex analysis gets the deeper reasoning it needs. Because the rate scales with the mode you choose, your agent hour usage reflects both how long Quick works and which mode you run it in.

Often, yes. A request in Quick may involve several steps, and some of those steps call out to an external agent or tool. Many of those calls are made by the agent and synchronous by design, meaning Quick has to stay engaged and wait for the result before it can continue. That waiting is required for Quick to finish your request, so the time generally counts against your agent hours. For example, if you ask Quick to update an opportunity in Salesforce, it holds the connection open until the write completes, and that time counts because Quick is actively working the request rather than sitting idle. Since many integrations work this way, in most cases you can expect this time to be included in your usage.

Quick maintains the memory and context of conversations in several ways, and Quick uses caching to optimize response quality, accuracy, performance, and cost automatically for users. Loading older chats to review previous outputs and conversation history does not use agent hours.

Monthly allowances:

Note: BI Author licenses, available on Enterprise accounts, receive a free upgrade to Enterprise licenses until 12/31/2026. The same allowances apply.

It is not currently possible to enter overages in the Professional Plan. Enterprise users can exceed their monthly entitlement. Overage is priced at $3 per agent hour and $5 per GB of index storage. Administrators can deploy overage limit policies. Index is pooled across users in an account, both entitlement as well as overages. For Agent Hours, only overages are pooled.

As an account administrator, you can manage overages using Limit Profiles. A limit profile can be applied to any user or role, or set as a default for the whole account. Consumption is evaluated in real time, and usage is drawn from remaining entitlements first. After all entitlements are exhausted, overage begins but does not exceed the policy set by the administrator. Limit policies can be applied to both agent hours and index storage.

For users on the Free, Plus, or Max plans, increase agent hours by upgrading the pricing plan you are subscribed to, or by enabling overage in the account and billing settings page. For Professional and Enterprise plan holders, there is currently no in-product mechanism to request a limits increase through approval workflows.

You can see your agent hour and index usage under Mission Control on the Activity page. If your account administrator has applied a limits policy, the widget reflects usage against the total: your entitlement plus the limits set in the policy.

Administrators have several ways to monitor usage. In the Admin console, you can open the Index Dashboard to look up consumption of both index storage and agent hours, broken down per user.

Index storage is measured in GB of raw storage — the original file size of documents at their source location (such as SharePoint, S3, or Confluence) before any processing by Amazon Quick. Index storage is allotted per user and pooled at the organizational level.

Storage allowances by plan:

Index storage overages are charged at $5 per GB per month beyond the total pooled allocation for an organization.

Professional and Enterprise accounts are charged a flat $250 per account per month infrastructure fee. This fee covers the underlying AI infrastructure that powers the Amazon Quick experience.

Amazon Quick offers a free trial option for new customers wanting to try the Enterprise plan:

During the 30-day free trial period:

• The full subscription fee is waived for up to 25 users
• The $250 infrastructure fee is waived for the account
• Customers can access all Enterprise Features. At the end of 30 days, customers with an AWS account will start being billed for an Enterprise plan.
• No credit card required to begin

Start your 30-day free trial

In addition to the core Quick subscription, customers can unlock advanced BI capabilities for Amazon Quick Sight:

SPICE — In-Memory Calculation Engine

An optional in-memory calculation engine for Amazon Quick Sight that dramatically accelerates query performance on large datasets.

• Price: $0.38 / GB / month
• Best for: Teams running frequent or complex analytics queries who want fast, responsive dashboards

Pixel-Perfect Reports

For users who create and share highly formatted, multi-page reports and schedule data exports at scale.

• Price: $1 / report unit / month (minimum 500 report units per month)
• Best for: Finance, operations, and compliance teams producing standardized recurring reports

Alerts and Anomaly Detection

Amazon Quick Sight supports both threshold-based alerts and automatic ML-powered anomaly detection that notify users of important changes in their data.

• Price: Based on number of metrics evaluated
• Types: Threshold-based alerts | ML anomaly detection (auto)
• Best for: Operations, finance, and data teams monitoring KPIs and business metrics

BI-Only Pricing Option: Customers using Quick exclusively for BI can access dedicated BI-only pricing options. Learn more about Quick Sight BI-only pricing.