deployment

# Claude Code Deployment

## Overview

Claude Code supports deployment through multiple providers beyond the direct Claude API, including Amazon Bedrock and Google Vertex AI for enterprise cloud deployment.

## Amazon Bedrock Integration

### Overview
Claude Code integrates with Amazon Bedrock to enable deployment through AWS infrastructure using Claude models available in your AWS account.

### Prerequisites
- Active AWS account with Bedrock access enabled
- Access to desired Claude models (e.g., Claude Sonnet 4.5)
- AWS CLI installed (optional)
- Appropriate IAM permissions

### Setup Process

#### 1. Model Access
Navigate to the Amazon Bedrock console, access Model access settings, and request Claude model availability in your region.

#### 2. AWS Credentials Configuration
Multiple authentication methods are supported:

**AWS CLI:**
```bash
aws configure
```

**Environment variables:**
```bash
export AWS_ACCESS_KEY_ID=your-key
export AWS_SECRET_ACCESS_KEY=your-secret
export AWS_SESSION_TOKEN=your-token  # Optional
```

**SSO profile:**
```bash
aws sso login --profile=<name>
export AWS_PROFILE=your-profile
```

**Bedrock API keys:**
```bash
export AWS_BEARER_TOKEN_BEDROCK=your-token
```

#### 3. Claude Code Configuration
Enable Bedrock integration:
```bash
export CLAUDE_CODE_USE_BEDROCK=1
export AWS_REGION=us-east-1  # Or preferred region
```

Optional override for Haiku region:
```bash
export ANTHROPIC_SMALL_FAST_MODEL_AWS_REGION=us-west-2
```

#### 4. Model Selection
Default models include Claude Sonnet 4.5 and Claude Haiku 4.5.

Customize via:
```bash
export ANTHROPIC_MODEL='model-id'
export ANTHROPIC_SMALL_FAST_MODEL='haiku-model-id'
```

#### 5. Token Configuration
Recommended settings:
```bash
export CLAUDE_CODE_MAX_OUTPUT_TOKENS=4096
export MAX_THINKING_TOKENS=1024
```

### IAM Permissions

Required actions:
- `bedrock:InvokeModel`
- `bedrock:InvokeModelWithResponseStream`
- `bedrock:ListInferenceProfiles`

Example IAM policy:
```json
{
  "Version": "2012-1