For enterprise teams using AWS Bedrock as their AI provider, tracking which teams or applications are consuming AI resources can be difficult. AWS bills Bedrock usage at the account level by default, making it hard to attribute costs to specific projects, departments, or use cases.
elvex supports AWS Bedrock application inference profiles, which let you route your AI usage through a profile ARN for granular cost attribution and usage tracking directly in your AWS account.
What are inference profiles?
An AWS Bedrock application inference profile is a configuration that wraps a foundation model and adds metadata so AWS can attribute usage to specific sources. When you configure elvex to use an inference profile ARN, all AI calls from elvex are routed through that profile, and the usage appears tagged in your AWS Cost Explorer.
You might use inference profiles when:
Your organization needs to track AI spend by team, department, or project in AWS
You have compliance requirements around AI usage attribution
You want to set usage limits or budgets per application in AWS
Your AWS administrator has set up inference profiles as the standard way to access Bedrock models
How to configure a Bedrock provider with an inference profile ARN
Navigate to Settings → AI Providers as an admin
Find your AWS Bedrock provider and click to edit it, or click + Add Provider to create a new one
In the model configuration field, paste your application inference profile ARN (e.g.
arn:aws:bedrock:us-east-1:123456789012:application-inference-profile/my-profile)The form will automatically detect that you've entered a profile ARN and display an Inference profile chip
It will also show the detected AWS region from the ARN
Complete any other required fields and save the provider
elvex will use the profile ARN when invoking Bedrock, and will automatically resolve the underlying foundation model for token limits and other settings.
Limitations and considerations
Application inference profiles must be created in your AWS account before configuring them in elvex — elvex does not create profiles for you
The profile must grant access to a supported Bedrock foundation model
This feature is intended for enterprise AWS customers managing Bedrock at scale; most teams using Bedrock directly with a model ID do not need to change anything
Common Questions
Do I need to change anything if I'm already using Bedrock with a model ID?
No — existing Bedrock configurations using model IDs continue to work exactly as before. Inference profile support is an additional option for teams that need it.
Where do I create an application inference profile?
Inference profiles are created in the AWS Bedrock console or via the AWS CLI/SDK. Contact your AWS administrator if you need help setting one up.
Will my token limits and model settings still work with a profile ARN?
Yes — elvex automatically resolves the underlying foundation model from the profile ARN and applies the correct token limits, temperature settings, and other model-specific configurations.