Claude Sonnet 4 on Bedrock - 2 requests/minute rate limit clarification
I'm building a development tool that integrates AI coding assistants with AWS Bedrock. I'm hitting severe rate limits with Claude Sonnet 4:
- Cross-region requests: 2/minute (L-559DCC33)
- Global requests: 1/minute (L-C63AA5DA)
- Both marked "Not Adjustable"
This makes interactive development impossible (Claude Code initialization alone uses 2-3 calls).
Questions:
- Is 2 req/min the intended limit for on-demand Sonnet 4?
- Are there architectural patterns to work within these constraints?
- What's the recommended path for development use cases?
Use case: CLI tool provisioning Bedrock endpoints for AI coding assistants.
- Language
- English
asked a year ago985 views
- Newest
- Most votes
- Most comments
Yes, the low request rate limits you're encountering with Claude 3.5 Sonnet on Amazon Bedrock are the default quotas set for the model. These limits are indeed very restrictive for interactive development use cases like yours.
To address your specific questions:
-
The default quota for on-demand InvokeModel requests per minute for Claude 3.5 Sonnet is indeed very low (1-2 requests per minute), which aligns with what you're experiencing. This is the intended default limit for on-demand usage.
-
For architectural patterns within these constraints, you have a few options:
- Implement retry logic with exponential backoff in your application
- Make your prompts more concise to reduce token usage
- Space out your requests to stay within the per-minute limits
- Consider a multi-account strategy as a temporary solution (since limits are at the account level)
- Monitor your usage with CloudWatch metrics to better understand your consumption patterns
- For development use cases, the recommended path is to:
- Contact AWS Support directly to request a quota increase for "On-demand InvokeModel requests per minute" for the Claude model
- Clearly explain your development use case and why you need higher quotas
- Provide details about your current usage and the performance issues you're facing
- Specify the desired quota increase you're requesting
While the Service Quotas console shows these limits as "not adjustable," you can still request increases through AWS Support. However, approval isn't guaranteed and may take some time. AWS typically prioritizes accounts that are already utilizing a high percentage of their current quota.
For high-throughput requirements, you might also consider exploring Provisioned Throughput options, though this comes with additional costs.
Sources
Why cant i invoke claude 3.5 more than one per minute ? | AWS re:Post
Building Agentic AI app with Strands Agents & MCP | AWS re:Post
How to increase request rate in AWS Bedrock | AWS re:Post
Bedrock Test Agent quota error | AWS re:Post
answered a year ago
Relevant content
asked a year ago
asked a year ago
