Making inference requests - Amazon Bedrock
Services or capabilities described in AWS documentation might vary by Region. To see the differences applicable to the AWS European Sovereign Cloud Region, see the AWS European Sovereign Cloud User Guide.

Making inference requests

Inference is the process of generating an output from an input provided to a model. Before you can send an inference request to Amazon Bedrock, you need to allow your role to perform the model invocation API actions. This depends on the endpoint you are using.

bedrock-mantle endpoint

If your role has the AmazonBedrockMantleInferenceAccess AWS managed policy attached, you can skip this section. Otherwise, attach the following permissions to allow inference via the bedrock-mantle endpoint (Responses API, Chat Completions, Messages API):

{
    "Version": "2012-10-17", 
    "Statement": [
        {
            "Sid": "MantleInferencePermissions",
            "Effect": "Allow",
            "Action": [
                "bedrock-mantle:CreateInference",
                "bedrock-mantle:GetProject",
                "bedrock-mantle:ListProjects",
                "bedrock-mantle:ListTagsForResources"
            ],
            "Resource": "*"
        }
    ]
}

bedrock-runtime endpoint

If your role has the AmazonBedrockFullAccess AWS managed policy attached, you can skip this section. Otherwise, attach the following permissions to allow inference via the bedrock-runtime endpoint (Converse and Invoke APIs):

{
    "Version": "2012-10-17", 
    "Statement": [
        {
            "Sid": "ModelInvocationPermissions",
            "Effect": "Allow",
            "Action": [
                "bedrock:InvokeModel",
                "bedrock:InvokeModelWithResponseStream",
                "bedrock:GetInferenceProfile",
                "bedrock:ListInferenceProfiles",
                "bedrock:RenderPrompt",
                "bedrock:GetCustomModel",
                "bedrock:ListCustomModels",
                "bedrock:GetImportedModel",
                "bedrock:ListImportedModels",
                "bedrock:GetProvisionedModelThroughput",
                "bedrock:ListProvisionedModelThroughputs",
                "bedrock:GetGuardrail",
                "bedrock:ListGuardrails",
                "bedrock:ApplyGuardrail"
            ],
            "Resource": "*"
        }
    ]
}

For a detailed breakdown of each permission, see Prerequisites for running model inference.

Topics