- Newest
- Most votes
- Most comments
Hello.
Is the bedrock-mantle endpoint quota system intentionally starting new accounts at 0? Is this a known issue?
Is there a way to self-serve reset/restore quotas to the AWS default values without waiting for a support case?
If you have already contacted AWS support, all you can do is wait for their response.
Quota issues can only be resolved by AWS support.
https://repost.aws/questions/QU3vGNsL2TS1OTkcWjDwcN9w/amazon-bedrock-throttlingexception-on-all-models-account-shows-0-applied-quota#ANX9W0SqoGSTe21ZqGGBPPmQ
Why would the bedrock-mantle quota system override/ignore the standard AWS default values?
Are other AWS Activate accounts seeing this same issue?
I often see this happening with recently created AWS accounts.
Based on the responses to other posts, it seems to be a measure to prevent exorbitant charges from being incurred.
https://repost.aws/questions/QUf16LkLwNS2yRbKVufdu6FA/new-account-stuck-at-0-tpm-rpm-for-all-bedrock-models-despite-aws-default-showing-5-000-000-tpm-provisioning-issue
I understand you're experiencing a critical production outage with Amazon Bedrock. Based on your description, this appears to be related to the bedrock-mantle endpoint quota system showing 0 for applied account-level quotas while AWS defaults show 5,000,000 TPM.
Key points about this situation:
Understanding the quota system: The bedrock-mantle endpoint represents cross-region inference profiles in Amazon Bedrock, which allow you to route requests across multiple AWS regions for higher throughput and resilience. These profiles have separate quota management from standard single-region endpoints.
Why quotas might show as 0: When cross-region inference profiles are newly available or when accounts are first accessing them, the applied quotas may not automatically inherit the default values. This is particularly common with newer features or endpoint types. The 0 value indicates that no quota has been explicitly allocated to your account for these specific endpoints yet, even though defaults exist.
Immediate actions you should take:
-
Verify you're using the correct endpoint: If you're trying to use cross-region inference profiles (bedrock-mantle endpoints) but haven't explicitly requested access to them, switch to standard regional endpoints temporarily. Standard Bedrock endpoints in us-east-1 should have normal quota allocations.
-
Check your quota increase requests: Since you've submitted 8 quota increase requests, these need to be processed by AWS support. The 24-hour timeframe, while frustrating, is not unusual for quota increases, especially for newer endpoint types.
-
Escalate through proper channels: As an AWS Activate startup, you should have access to startup support resources. Contact your AWS Activate team directly or use your AWS account team if you have one assigned.
-
Temporary workaround: If cross-region profiles aren't critical for your use case, explicitly configure your application to use standard regional Bedrock endpoints (not the bedrock-mantle cross-region profiles) until your quota increases are approved.
Why this is happening: Cross-region inference profiles are a relatively newer feature, and their quota management is separate from standard Bedrock endpoints. Accounts may need explicit quota allocation for these endpoints even when standard Bedrock quotas are available.
This is not a standard behavior for regular Bedrock endpoints and appears specific to the cross-region inference profile system. Your support cases are the correct path forward, but in the meantime, switching to standard regional endpoints should restore service.
Worth taking a look at this as well:
https://docs.aws.amazon.com/bedrock/latest/userguide/quotas.html
Hey, we are also under AWS Startup activate program. Same issue is happening with us. Try using those models in Playground if they are accessible or not. I think Claude models are currently having issues as they need more compliance. Try using any other model like Nova2 lite in playground if they are accessible or not. There might be payment issue also check if you have autopay enabled or not in payment prefernece in billing and cost managment.
answered 2 months ago
Relevant content
asked 3 months ago
- AWS OFFICIALUpdated 10 months ago
- AWS OFFICIALUpdated a year ago
