AWS Cloud Financial Management

AWS Price List Gets a Natural Language Upgrade: Introducing the AWS Pricing MCP Server

Transform complex pricing, availability, and service information into simple conversations with your AI assistant

Have you ever wanted to quickly understand AWS pricing without navigating through multiple pricing pages or writing code? As AWS’s service portfolio expands to over 240+ services across 43+ regions, cloud professionals face a growing challenge: get product, regional availability, and pricing information faster and make timely decisions. We believe the future of interacting with AWS product and pricing information lies in natural language conversations directly within your existing tools and development environments, eliminating the need to toggle between documentation and enabling seamless decision-making.

Today, that changes. We are excited to release the aws-pricing-mcp-server, an open-source tool in the AWS Labs GitHub repository, that brings natural language pricing queries to your favorite AI assistants through the Model Context Protocol (MCP). Now you can simply ask “What would it cost to run three m5.large instances and a MySQL RDS database in us-west-2?” and get instant pricing answers using natural language queries without leaving your workflow.

How Model Context Protocol simplifies workload pricing

The Model Context Protocol (MCP) provides a standardized way for AI agents to connect to real-time data sources. Think of it as a universal translator that allows your AI assistant to access current AWS pricing data directly, eliminating guesswork and manual calculations.

How the AWS Pricing MCP Server works

The aws-pricing-mcp-server uses MCP to create a direct bridge between your AI assistants and the AWS Price List API. This allows your AI assistant like Amazon Q to access real-time pricing data. For example, when using Amazon Q Developer Command Line Interface (CLI) with the aws-pricing-mcp-server, you can ask questions about AWS services in natural language or provide infrastructure code like CDK or Terraform files to estimate costs. Amazon Q will automatically:

  1. Parse your requirements (services, instance types, regions, availability needs)
  2. Query the AWS Price List API for current and historical pricing, product availability, regional availability, and product attribute information.
  3. Process complex multi-service information including pricing comparisons, regional availability, and service specifications
  4. Present comprehensive results in an easy-to-understand format that helps you make informed decisions.

Why use it?

Natural language access to AWS service and pricing information benefits multiple job roles:

  • For Developers: Stay in your development flow without context-switching to pricing calculators or documentation. When evaluating architectural choices, ask about cost implications directly from your terminal or IDE. Questions like “How much does it cost to create an Amazon Connect instance?” receive immediate, data-driven answers that help you build cost-optimized solutions.
  • For FinOps: Transform complex resource planning into simple conversations. Instead of creating spreadsheets and navigating multiple pricing pages, quickly identify the ideal resources and regions for new workloads with questions like “Compare the cost of processing 2 million credit card transactions using AWS KMS between us-east-1 and eu-west-1, assuming each transaction requires 2 API calls.”
  • For Solution Architects: Make informed decisions about regional deployments and service compatibility. Ask specific questions about service availability like “Which regions support AWS Wavelength?” to quickly evaluate infrastructure options across AWS’s global footprint.
  • For Cloud Governance Teams: Streamline compliance and governance by quickly identifying regional service differences with queries like “Does S3 Object Lock support legal hold in all regions?”

See it in action

Let us walk through two scenarios that demonstrate the power of natural language queries for AWS services, pricing, and availability. The following examples were generated using Amazon Q CLI and the aws-pricing-mcp-server.

Example 1: “What is the lowest price for EC2 instances in Dublin, that have at least 7 GB of RAM and 30 GB of local SSD storage?”

Response:

Based on my analysis of EC2 pricing in Dublin (EU Ireland), here are the lowest-cost instances that meet your requirements of at least 7 GB RAM and 30 GB local SSD storage:

Top 5 Most Cost-Effective Options:

1. r6gd.medium - $0.064/hour ⭐ CHEAPEST
• Memory: 8 GiB
• Storage: 1 x 59 NVMe SSD
• vCPUs: 1
• Monthly cost: ~$46.08

2. r7gd.medium - $0.0756/hour
• Memory: 8 GiB
• Storage: 1 x 59 NVMe SSD
• vCPUs: 1
• Monthly cost: ~$54.43

3. x2gd.medium - $0.10/hour
• Memory: 16 GiB
• Storage: 1 x 59 SSD
• vCPUs: 1
• Monthly cost: ~$72.00

4. m6gd.large - $0.1008/hour
• Memory: 8 GiB
• Storage: 1 x 118 NVMe SSD
• vCPUs: 2
• Monthly cost: ~$72.58

5. m5ad.large - $0.115/hour
• Memory: 8 GiB
• Storage: 1 x 75 NVMe SSD
• vCPUs: 2
• Monthly cost: ~$82.80

Key Findings:

• Absolute lowest cost: r6gd.medium at $0.064/hour ($46.08/month)
• Best value for more storage: m6gd.large at $0.1008/hour with 118 GB NVMe SSD
• Best value for more memory: x2gd.medium at $0.10/hour with 16 GiB RAM

All prices are for On-Demand Linux instances with shared tenancy. The r6gd.medium offers the best price point while exceeding your minimum requirements with 8 GB RAM and 59 GB NVMe SSD storage.

Example 2: “I’m building a messaging system. Compare the pricing models for Amazon SQS (Simple Queue Service) Standard queues versus Amazon SQS FIFO queues for a workload involving 100 million messages per month. What are the key functional differences and when should I choose one over the other?”

Response:

Cost Comparison (100M messages/month)

| Queue Type     | Price per Million Requests | Monthly Cost (after free tier) |
|----------------|----------------------------|--------------------------------|
| Standard Queue | $0.40                      | $39.60                         |
| FIFO Queue     | $0.50                      | $49.50                         |

Cost Difference: FIFO queues cost 25% more than Standard queues ($0.50 vs $0.40 per million requests).

Key Functional Differences

Standard Queues
• Message Delivery: At least once delivery (may occasionally deliver duplicates)
• Message Order: Best-effort ordering (not guaranteed)
• Throughput: Nearly unlimited throughput
• Use Cases: High-throughput applications where occasional duplicates and out-of-order processing are acceptable

FIFO Queues
• Message Delivery: Exactly-once processing (no duplicates)
• Message Order: Strict first-in, first-out delivery guaranteed
• Throughput: Limited to 300 transactions per second (TPS) by default, up to 3,000 TPS with high throughput mode
• Message Groups: Support for message grouping (messages with same group ID processed in order)
• Deduplication: Built-in content-based deduplication or developer-provided deduplication IDs

When to Choose Each Queue Type

Choose Standard Queues when:
• Maximum throughput is critical
• Your application can handle occasional duplicate messages
• Strict ordering isn't required
• Cost efficiency is a priority
• You need to process millions of messages per second

Choose FIFO Queues when:
• Exact ordering of messages is critical (e.g., financial transactions, order processing)
• You cannot tolerate duplicate messages
• Your throughput requirements are under 3,000 TPS
• You need message grouping functionality
• You're willing to pay a 25% premium for these guarantees

Recommendations for Your Messaging System

1. Assess Your Requirements:
• Do you need guaranteed ordering?
• Is duplicate processing problematic for your use case?
• What's your expected throughput?

2. Consider a Hybrid Approach:
• Use FIFO queues for critical, order-dependent operations
• Use Standard queues for high-volume, less sensitive operations

3. Cost Optimization:
• Batch messages when possible (up to 10 messages per API call)
• Implement client-side deduplication for Standard queues if needed
• Use long polling to reduce the number of empty receives

For your specific workload of 100 million messages per month, the cost difference is approximately $10/month ($49.50 vs $39.60). This relatively small difference means your decision should be primarily based on your functional requirements rather than
cost considerations.

Getting Started

The aws-pricing-mcp-server integrates seamlessly with popular AI assistants including Amazon Q Developer CLI, Claude Desktop, Kiro, and other MCP compatible tools.

Prerequisites

  1. Install uv from Astral or the GitHub README
  2. Install Python using uv python install 3.10
  3. Set up AWS credentials with access to AWS services
    1. You need an AWS account with appropriate permissions
    2. Configure AWS credentials with aws configure or environment variables
    3. Ensure your IAM role/user has permissions to access AWS Price List API
  4. An MCP compatible AI assistant (e.g., Q for CLI, Claude Desktop, Kiro)

Installation

Quick Setup with Amazon Q Developer CLI
Configure the MCP server in your MCP client configuration (For e.g., for Amazon Q Developer CLI add the servers to your ~/.aws/amazonq/mcp.json file)

{
  "mcpServers": {
    "awslabs.aws-pricing-mcp-server": {
      "command": "uvx",
      "args": ["awslabs.aws-pricing-mcp-server@latest"],
      "env": {
        "FASTMCP_LOG_LEVEL": "ERROR",
        "AWS_PROFILE": "your-aws-profile"
      },
      "disabled": false,
      "autoApprove": []
    }
  }
}

Docker Deployment for Teams
Organizations with multiple users across engineering and finance teams can leverage Docker for reproducible, standardized deployment:

Build the Docker image:

git clone https://github.com/awslabs/mcp
cd mcp/src/aws-pricing-mcp-server
docker build -t aws-pricing-mcp-server .

Create a .env file with your AWS credentials:

AWS_ACCESS_KEY_ID=your_access_key
AWS_SECRET_ACCESS_KEY=your_secret_key
AWS_REGION=your_region

NOTE: Your credentials will need to be kept refreshed from your host

Configure the MCP server in your MCP client configuration (for example, for Amazon Q Developer CLI, add the servers to your ~/.aws/amazonq/mcp.json file):

{
  "mcpServers": {
    "awslabs.aws-pricing-mcp-server": {
      "command": "docker",
      "args": [
        "run",
        "--rm",
        "--interactive",
        "--env",
        "FASTMCP_LOG_LEVEL=ERROR",
        "--env-file",
        "/full/path/to/file/above/.env",
        "awslabs/aws-pricing-mcp-server:latest"
      ],
      "env": {},
      "disabled": false,
      "autoApprove": []
    }
  }
}

Authentication

The MCP server requires specific AWS permissions and configuration:

Required Permissions

Your AWS IAM role, or user, must have pricing:* permissions to access the AWS Pricing API. The server accesses generally available AWS pricing information and does not retrieve any user-specific data. All pricing API calls are free of charge and do not incur any costs.

Configuration

The server uses two key environment variables:

  • AWS_PROFILE: Specifies the AWS profile to use from your AWS configuration file. If not provided, it defaults to the “default” profile.
  • AWS_REGION: Determines the geographically closest AWS Pricing API endpoint to use. This improves performance by routing requests to the nearest regional endpoint.
"env": {
  "AWS_PROFILE": "your-aws-profile",
  "AWS_REGION": "us-east-1"
}

Conclusion

The aws-pricing-mcp-server offers a new way to interact with AWS pricing and service information. By simplifying access to this data through natural language, it provides an additional option for accessing AWS service information directly within your existing workflows.

Ready to try it out? Visit the AWS Labs GitHub repository for documentation, usage examples, and support resources. Whether you are a business analyst planning next quarter’s cloud budget, an architect evaluating service availability across regions, or a developer making real-time architectural decisions, natural language queries can help streamline your workflow and improve your decision-making process.

Try the aws-pricing-mcp-server today and discover how natural language can transform your relationship with AWS services.

Fanita Rodrigues

Fanita Rodrigues

Senior Technical Product Manager at Amazon Web Services, focused on AWS Pricing Experiences, including the AWS Price List and related customer tools. She leads initiatives that improve how customers discover, understand, and integrate AWS pricing into their workflows. Prior to AWS, Fanita delivered solutions to improve Amazon’s International Return Experience for customers. Outside of work, she’s an avid traveler with a special love for national parks, and a home chef who enjoys exploring global cuisines and recreating them in her kitchen.

Christian Busch

Christian Busch

Software Development Manager at Amazon Web Services. Leading the Price List team, he is driving how millions of customers discover products and prices by building and evolving scalable, distributed systems. He has a passion for high-quality software and enjoys being at the forefront of technology in topics like artificial intelligence and high performance computing. Christian holds a university diploma in Computer Science from the Karlsruhe Institute of Technology. Outside of work, he enjoys reading computer science journals, building a RISC V CPU and running half marathons.

Puneet Jindal

Puneet Jindal

Senior Manager of Technical Product Management at Amazon Web Services, leading product strategy and delivery for AWS pricing experiences. He drives how millions of customers price, evaluate and procure AWS services globally by building scalable platforms that simplify cloud pricing. Prior to AWS, Puneet delivered solutions for cloud infrastructure security and Amazon logistics. Outside of work, Puneet enjoys Formula 1 racing and building home automation projects.