Can I use Stockyard with the OpenAI Python SDK?

Yes. Set the base_url to your Stockyard endpoint and use your existing API key. All OpenAI SDK features including streaming, function calling, and vision work through the proxy.

Does Stockyard cache OpenAI responses?

Yes. The built-in prompt cache stores responses for identical requests. Cache hits are free and return in single-digit milliseconds instead of waiting 1-30 seconds for the model.

Can I fail over from OpenAI to Anthropic?

Yes. Configure both providers and set up model aliasing. When OpenAI returns errors or times out, Stockyard automatically retries on your fallback provider.

OpenAI Proxy — Route GPT-4o Through Stockyard

Environment variable

OPENAI_API_KEY

Models

gpt-4o, gpt-4o-mini, gpt-4.1, gpt-4.1-mini, o3-mini

Failover to

Anthropic Claude, Google Gemini, or Groq

API format

OpenAI-compatible

Why proxy OpenAI?

OpenAI is the most-used LLM provider but gives you limited visibility into costs until your bill arrives. Proxying through Stockyard adds per-request cost tracking, response caching (identical prompts skip the API entirely), rate limiting per user or project, and automatic failover to Anthropic or other providers when OpenAI has an outage.

Your application code does not change. Stockyard speaks the same OpenAI-compatible API, so any SDK that works with OpenAI works with Stockyard.

Quick start

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

# Set your OpenAI API key
export OPENAI_API_KEY=your-key-here

# Start the proxy
stockyard
# Provider: openai (from OPENAI_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":"gpt-4o","messages":[{"role":"user","content":"hello"}]}'

Good to know

OpenAI models are auto-detected from the model name in the request. No extra config needed beyond the API key.

Direct OpenAI vs through Stockyard

DIRECT TO OPENAI

Requests go straight to api.openai.com. No cost visibility until the monthly invoice. No caching. No failover. No audit trail.

THROUGH STOCKYARD

Every request logged with cost. Identical prompts cached automatically. If OpenAI goes down, failover to Claude or Gemini. Hash-chained audit trail for compliance.

Route OpenAI through Stockyard

Why proxy OpenAI?

Quick start

Good to know

Direct OpenAI vs through Stockyard