Skip to main content

​
Overview

Amazon Bedrock is a fully managed service that provides access to foundation models from leading AI providers — including Anthropic Claude, Meta Llama, and Amazon Titan — through a unified AWS API. Bedrock integrates with the AWS ecosystem for security, networking, and billing, making it ideal for organizations already invested in AWS infrastructure. Key benefits:
  • Multi-provider access — Use Claude, Llama, Titan, and other models through a single API
  • AWS integration — IAM authentication, VPC endpoints, CloudWatch monitoring, and CloudTrail auditing
  • No infrastructure management — Fully managed serverless inference with automatic scaling
  • Data privacy — Your data is not used to train foundation models

​
Setup

1

Enable Bedrock Access

In the AWS Console, navigate to Amazon Bedrock and enable access to the foundation models you want to use. Model access must be requested per-model and per-region.
Some models (e.g., Anthropic Claude) require you to accept usage terms before access is granted. This can take a few minutes to process.
2

Create IAM Credentials

Create an IAM user or role with the bedrock:InvokeModel and bedrock:InvokeModelWithResponseStream permissions. For programmatic access, generate an access key.Minimal IAM policy:
{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Effect": "Allow",
      "Action": [
        "bedrock:InvokeModel",
        "bedrock:InvokeModelWithResponseStream",
        "bedrock:ListFoundationModels"
      ],
      "Resource": "*"
    }
  ]
}
3

Configure in Nadoo

Go to Admin > Model Providers > AWS Bedrock and enter:
FieldRequiredDescription
AWS RegionYesThe AWS region where Bedrock is enabled (e.g., us-east-1)
Access Key IDYes*AWS access key for authentication
Secret Access KeyYes*AWS secret key for authentication
*If running on AWS infrastructure (EC2, ECS, Lambda), you can use an IAM role instead of explicit keys.
4

Test Connection

Click Test to verify the credentials and region are valid. Enable the models you want to use.

​
Available Models

​
Anthropic Claude (via Bedrock)

Model IDContext WindowBest For
anthropic.claude-3-5-sonnet-20241022-v2:0200K tokensBest balance of intelligence and speed
anthropic.claude-3-opus-20240229-v1:0200K tokensHighest reasoning capability
anthropic.claude-3-haiku-20240307-v1:0200K tokensFastest and most cost-efficient

​
Meta Llama

Model IDContext WindowBest For
meta.llama3-1-70b-instruct-v1:0128K tokensStrong open-source alternative
meta.llama3-1-8b-instruct-v1:0128K tokensCost-efficient smaller model

​
Amazon Titan

Model IDContext WindowBest For
amazon.titan-text-premier-v1:032K tokensAWS-native text generation
amazon.titan-embed-text-v2:08K tokensEmbeddings for RAG
Model availability varies by AWS region. Check the Bedrock model availability page for the latest supported regions per model.

​
Capabilities

Chat Completion

Conversational AI with streaming support across all Bedrock LLM models.

Embeddings

Generate vector embeddings using Amazon Titan Embed for knowledge base indexing and semantic search.

AWS Security

IAM-based access control, VPC endpoints for private connectivity, and CloudTrail auditing.

Multi-Provider

Access Claude, Llama, Titan, and other models through a single provider configuration.

​
Authentication Methods

Nadoo AI supports two authentication methods for AWS Bedrock.

​
Access Keys

Provide an AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY directly. Suitable for self-hosted deployments outside AWS. When running Nadoo AI on AWS infrastructure (EC2, ECS, EKS, Lambda), attach an IAM role with Bedrock permissions. No explicit keys are needed — credentials are obtained automatically from the instance metadata service.
IAM roles are the recommended authentication method when running on AWS. They eliminate the need to manage and rotate access keys.

​
Environment Variables

When self-hosting, configure AWS Bedrock via environment variables: 
AWS_ACCESS_KEY_ID=AKIAIOSFODNN7EXAMPLE
AWS_SECRET_ACCESS_KEY=wJalrXUtnFEMI/K7MDENG/bPxRfiCYEXAMPLEKEY
AWS_REGION=us-east-1
If both environment variables and the admin UI configuration are set, the admin UI values take precedence.
Use CaseRecommended ModelReason
General chatbotClaude 3.5 SonnetBest all-around performance
Complex reasoningClaude 3 OpusHighest reasoning capability
High-volume supportClaude 3 HaikuFast and cost-efficient
Open-source preferenceLlama 3.1 70BStrong performance without proprietary models
Knowledge base searchTitan Embed Text v2AWS-native embedding model
Budget-constrainedLlama 3.1 8BLowest cost per token

​
Cross-Region Inference

Bedrock supports cross-region inference profiles that route requests to the nearest available region for lower latency and higher availability. To use cross-region inference, enable it in your Bedrock console and configure the inference profile ARN in Nadoo instead of a model ID.

​
Rate Limits and Pricing

Bedrock uses on-demand pricing with per-token charges that vary by model. You can also purchase Provisioned Throughput for guaranteed capacity at a fixed hourly rate.
Bedrock applies default quotas on tokens per minute and requests per minute. Check your account quotas in the AWS Service Quotas console and request increases before scaling production workloads.

​
Troubleshooting

Your IAM user or role does not have the required Bedrock permissions. Verify the IAM policy includes bedrock:InvokeModel and that you have enabled access to the specific model in the Bedrock console.
The model is not available in your selected region, or you have not requested access. Check the Bedrock console for model availability and access status.
You have exceeded your account’s request rate or token quota. Retry with exponential backoff or request a quota increase through AWS Service Quotas.
The request payload is malformed. Check that the model ID is correct and the request parameters match the model’s expected schema.