Route Anthropic through Stockyard

Why proxy Anthropic?

Anthropic uses a different API format than OpenAI. Stockyard translates between them transparently, so you can call Claude through the same /v1/chat/completions endpoint your app already uses for OpenAI. No SDK swap, no format conversion in your code.

This also means you can fail over between Claude and GPT-4o without changing your application. Set up model aliasing to route default-model to Claude, with GPT-4o as the backup.

Quick start

# Install Stockyard
curl -fsSL stockyard.dev/install.sh | sh

# Set your Anthropic API key
export ANTHROPIC_API_KEY=your-key-here

# Start the proxy
stockyard
# Provider: anthropic (from ANTHROPIC_API_KEY)
# Proxy listening on :4200

# Send a request through the proxy
curl http://localhost:4200/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{"model":"claude-sonnet-4-5-20250929","messages":[{"role":"user","content":"hello"}]}'

Good to know

Anthropic has different token counting and pricing than OpenAI. Stockyard normalizes both so your cost dashboard shows accurate per-request costs regardless of provider.

Use Claude with the OpenAI SDK

Anthropic has its own SDK and API format. With Stockyard, you don't need to use it. Send standard OpenAI-format requests and Stockyard translates them to Anthropic's format automatically:

# Use the OpenAI Python SDK to call Claude
from openai import OpenAI
client = OpenAI(base_url="http://localhost:4200/v1")

response = client.chat.completions.create(
    model="claude-sonnet-4-5-20250929",
    messages=[{"role": "user", "content": "hello"}]
)

This means you can switch between OpenAI and Anthropic models without changing your SDK, your error handling, or your response parsing. Just change the model name.