Amazon Bedrock refuses a Claude model ID with on-demand throughput
Invocation of model ID ... with on-demand throughput isn't supportedShort answer
Some Bedrock models, including newer Claude versions, are not served from a single Region by their bare model ID. You must call them through an inference profile: put the profile ID or ARN in modelId. List the profiles in your Region, use that ID, and make sure your IAM and Region policies allow every Region the profile routes to.
Fix it fast
1Find out which cause you have
List the system-defined inference profiles for your Region and look for your model. If it appears there, call that ID.
aws bedrock list-inference-profiles --region us-east-1 --type-equals SYSTEM_DEFINED \ --query "inferenceProfileSummaries[?contains(inferenceProfileId, 'claude')].[inferenceProfileId,inferenceProfileName]" \ --output table2Match what you see, then jump to its fix
- 1
The modelId in your code looks like anthropic.claude-... with no region prefixThe call uses the bare model ID instead of an inference profile IDFix - 2
The profile is not found, or is not available from the Region you callThe profile ID is for a different geography than your source RegionFix - 3
The profile ID is right but the call is deniedA policy blocks one of the Regions the profile routes toFix
- 1
What does the error look like?
Invocation of model ID ... with on-demand throughput isn't supported
Cause 1 of 3
The call uses the bare model ID instead of an inference profile ID
1Check
AWS documents that for a model served only through an inference profile you specify a cross-Region profile. Cross-Region profile IDs carry a geography prefix such as us. in front of the model ID.
grep -rn "anthropic.claude" . # find every modelId in your code2Fix
Replace the model ID with the ID (or ARN) of the inference profile from the list above, in InvokeModel and Converse alike.
import boto3
client = boto3.client("bedrock-runtime", region_name="us-east-1")
resp = client.converse(
modelId="us.anthropic.claude-3-haiku-20240307-v1:0", # example profile ID; use one from list-inference-profiles
messages=[{"role": "user", "content": [{"text": "Hello"}]}],
)3Verify
The call returns a reply instead of a validation error.
print(resp['output']['message']['content'][0]['text'])Confirmed by docs.aws.amazon.com
Cause 2 of 3
The profile ID is for a different geography than your source Region
1Check
Profiles are defined by the Regions they support, and available profiles differ by the source Region you call from. Ask Bedrock which profiles exist where you call.
aws bedrock get-inference-profile --region us-east-1 --inference-profile-identifier us.anthropic.claude-3-haiku-20240307-v1:0 --query 'models[].modelArn'2Fix
Take the ID from list-inference-profiles run in the same Region your client uses, or call from a Region the profile supports.
aws bedrock list-inference-profiles --region eu-west-1 --type-equals SYSTEM_DEFINED3Verify
get-inference-profile returns model ARNs for that Region.
aws bedrock get-inference-profile --region eu-west-1 --inference-profile-identifier <profile-id>Confirmed by docs.aws.amazon.com
Cause 3 of 3
A policy blocks one of the Regions the profile routes to
1Check
A request to a cross-Region profile can be routed to any destination Region in it. AWS states that if any destination Region is blocked in a service control policy, the request fails even when the others are allowed.
aws bedrock get-inference-profile --region us-east-1 --inference-profile-identifier <profile-id> --query 'models[].modelArn'2Fix
Allow the Bedrock inference actions (for example bedrock:InvokeModel*) in all destination Regions of the profile in your SCPs and IAM policies.
# IAM or SCP: allow in every destination Region listed above
"Action": ["bedrock:InvokeModel*"]3Verify
Repeat the call that failed.
print(client.converse(modelId='<profile-id>', messages=[{'role': 'user', 'content': [{'text': 'Hello'}]}])['output'])Confirmed by docs.aws.amazon.com
How do I stop it coming back?
- Keep the profile ID in config, not hard-coded, so a model upgrade is one edit.
- Read each model's page in the Bedrock docs for its inference profile IDs before you wire it in.
- Remember that inference profiles do not work with Provisioned Throughput.
Sources
How was this page made?
I drafted this page with AI assistance from the official documents listed under Sources. Every cause above names the document that confirms it. I have not reproduced this fix on a machine of my own yet, so read it as documented guidance, not a test report. Found a mistake? Tell me.
Which stack is behind this page?
Projects of mine built on the same technologies:
- Dispatch ProA real-time logistics platform: Flutter apps, Spring Boot services and a Kafka event pipeline on AWS.
- RaheebA consumer app shipped to both stores from one Flutter codebase on a .NET and Azure back end.
- Loade DriverThe driver side of a delivery platform: live jobs, routing and earnings, past 5K Google Play downloads.
Related fixes
AI
Azure OpenAI says the API deployment for this resource does not exist
The API deployment for this resource does not exist
Azure OpenAI does not take a model name in the model field.
2 causes, each with a fixAI
Azure OpenAI fails with an SSL certificate verification error
httpx.ConnectError: [SSL: CERTIFICATE_VERIFY_FAILED]
The client checks the endpoint's certificate against its CA bundle, which is certifi by default.
2 causes, each with a fixAI
Chroma data is empty after a restart in LangChain
The collection is empty after the app restarts
Chroma only keeps data when you ask it to.
3 causes, each with a fixAI
LangChain says it could not import the chromadb package
ImportError: Could not import chromadb python package. Please install it with `pip install chromadb`.
The message comes from LangChain's old Chroma class, which cannot import chromadb in the Python that runs your code.
2 causes, each with a fixAI
Gemini API returns 429 RESOURCE_EXHAUSTED on the free tier
429 RESOURCE_EXHAUSTED
Gemini limits are applied per Google Cloud project, not per API key, so a new key in the same project does not reset anything.
3 causes, each with a fixAI
Gemini API says the model is not found for gemini-1.5-pro or gemini-1.5-flash
models/gemini-1.5-flash is not found for API version v1beta
Google shut down gemini-1.5-pro, gemini-1.5-flash and gemini-1.5-flash-8b on 29 September 2025, so the API no longer finds them, whatever access you have.
3 causes, each with a fix