Service: Amazon Bedrock Category: Service Limit / Technical

The "Too many tokens per day" throttling error combined with Service Quotas showing an applied account-level quota value of 0 indicates that your account's token-based quotas for Amazon Bedrock model inference have been exhausted or restricted.

Here are the steps to address this issue:

Check Your Current Quota Usage: First, verify your actual quota limits and usage in the Service Quotas console. Navigate to Amazon Bedrock quotas and review the specific model inference quotas. A value of 0 may indicate that quotas haven't been properly configured or that you've reached your limit.

Request a Quota Increase: If your quotas are set too low or at 0, you'll need to request an increase through the Service Quotas console. Select the specific Bedrock model inference quota you need increased, and submit a request with justification for your use case. Be specific about which models you need access to and your expected token usage.

Verify Model Access: Ensure you have requested and been granted access to the specific foundation models you're trying to use. Model access must be explicitly requested through the Amazon Bedrock console for each model in each region where you plan to use it.

Review Your AWS Support Plan: The speed of quota increase requests can depend on your AWS Support plan. If you need faster resolution, consider reviewing your support plan options.

Add Details to Your Support Case: Since you have an existing case open, you can add additional correspondence through the AWS Support Center at https://console.aws.amazon.com/support/ with any additional details about your use case, expected usage patterns, and business impact to help expedite the review.

The quota restoration typically requires AWS to review and approve your request, especially if this affects all regions and models in your account.

Based on posts on re:Post, I have confirmed that your situation is caused by a known provisioning bug. According to reports from others who resolved this issue, AWS fixed it after they opened a case with Billing Support. Specifically, when submitting the case, they stated, "I am encountering a 'Too many requests' error while testing Bedrock agent models, even though my usage is very low," which led AWS to change their account quota from zero to the default value.

Bedrock: "Too many requests, please wait before trying again." | AWS re:Post

i have done everything and model access is not required for all