Route Cohere through Stockyard

Why proxy Cohere?

Cohere offers strong models for enterprise search, RAG, and classification. Their Command models are competitive on reasoning tasks and their Embed models are among the best for retrieval.

Proxying through Stockyard normalizes Cohere into the same OpenAI-compatible endpoint as your other providers. Track costs, cache responses, and fail over to other providers without changing your application code.

Quick start

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

# Set your Cohere API key
export COHERE_API_KEY=your-key-here

# Start the proxy
stockyard
# Provider: cohere (from COHERE_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":"command-r-plus","messages":[{"role":"user","content":"hello"}]}'

Good to know

Cohere uses their compatibility endpoint (/compatibility/v1) which Stockyard connects to automatically. Embedding requests route to Cohere when using embed model names.

Embeddings + chat through one proxy

Cohere's Embed models are among the best for RAG and semantic search. With Stockyard, you can route embedding requests to Cohere and chat requests to OpenAI or Anthropic through the same endpoint:

# Embeddings → Cohere
curl http://localhost:4200/v1/embeddings \
  -d '{"model":"embed-english-v3.0","input":"search query"}'

# Chat → OpenAI (same endpoint)
curl http://localhost:4200/v1/chat/completions \
  -d '{"model":"gpt-4o","messages":[...]}'

Both requests are traced, cost-tracked, and cached through the same proxy. One dashboard for everything.