Now supports MiniMax M2 & Claude Opus 4.5

One API for every
Bedrock model.

BedrockRouter is an OpenAI- and Anthropic-compatible proxy that runs entirely in your AWS account. Integrate Anthropic Claude, MiniMax, and every other Bedrock model through a single endpoint.

Browse modelsRead docs
Zero markup
Your account, your keys

Drop-in replacement

Switch to Bedrock without rewriting your stack.

If your code already uses the OpenAI or Anthropic SDK, you only need to change the base URL. Every feature works the same: streaming, tool use, vision, function calling.

  • Works with the OpenAI Python & Node SDKs
  • Works with the Anthropic Python & TypeScript SDKs
  • Works with Vercel AI SDK, LangChain, LlamaIndex, and more
app.py
from openai import OpenAI

client = OpenAI(
    base_url="https://YOUR_PROXY_URL/v1",
    api_key="YOUR_BEDROCK_KEY",
)

response = client.chat.completions.create(
    model="anthropic.claude-sonnet-4-5",
    messages=[{"role": "user", "content": "Hi!"}],
    stream=True,
)

for chunk in response:
    print(chunk.choices[0].delta.content, end="")

How it works

Production-ready in three steps.

01

Deploy CloudFormation

One click in the AWS console. The stack provisions a Lambda, IAM role, Function URL, and a Bedrock API key.

02

Copy your endpoint

Grab ProxyURL and ApiKey from the stack Outputs tab. That's your OpenAI-compatible base URL.

03

Ship

Point your existing OpenAI or Anthropic client at the proxy. Every request runs through your AWS account.

Why BedrockRouter

Built for teams that take security and cost seriously.

Runs in your AWS account

The proxy Lambda lives in your account. Requests never leave your infrastructure.

Native Bedrock API key

No SigV4, no IAM credentials. Single bearer token tied to your IAM permissions.

OpenAI & Anthropic compatible

Drop-in replacement for any OpenAI or Anthropic SDK. Works with Claude Code and Cursor.

Zero markup pricing

Pay AWS Bedrock's on-demand rates directly. We add nothing on top.

Full SSE streaming

Token-by-token streaming for both Anthropic and OpenAI response formats.

Instant deploy

One CloudFormation stack, two outputs, ready in 60 seconds.

Your account. Your keys. Your infrastructure.

Deploy in 60 seconds. Cancel any time by deleting the stack.