AWS Bedrock

Bedrock gives you Claude, Llama 3, Mistral, Titan, and Cohere models behind a single AWS contract and IAM boundary.

Requirements

Server-side (standard AWS credential chain — IAM role, env vars, or profile):

AWS_REGION=eu-west-1
AWS_ACCESS_KEY_ID=...
AWS_SECRET_ACCESS_KEY=...
# or use IAM Roles for Service Accounts (IRSA) in EKS

Option A — SDK

from tappass import Agent

agent = Agent("https://tappass.example.com", "tp_...")

# Claude via Bedrock
response = agent.chat(
    "Hello",
    model="bedrock/anthropic.claude-3-5-sonnet-20241022-v2:0",
)

# Llama via Bedrock
response = agent.chat(
    "Hello",
    model="bedrock/meta.llama3-1-70b-instruct-v1:0",
)

Option B — OpenAI SDK, zero-code

export OPENAI_BASE_URL=https://tappass.example.com/v1
export OPENAI_API_KEY=tp_...

from openai import OpenAI

client = OpenAI()
response = client.chat.completions.create(
    model="bedrock/anthropic.claude-3-5-sonnet-20241022-v2:0",
    messages=[{"role": "user", "content": "Hello"}],
)

What’s supported

• All Bedrock foundation models (Claude, Llama, Mistral, Titan, Cohere, AI21)
• Streaming + non-streaming via InvokeModelWithResponseStream
• Tool use (on models that support it — Claude, Llama 3.1+)
• Vision (Claude multimodal models)

Data residency

Use regional endpoints: eu-west-1 (Ireland), eu-central-1 (Frankfurt), eu-west-3 (Paris). Bedrock guarantees the model inference stays in-region.