Skip to main content

AutoGen

Microsoft AutoGen configures LLM access through llm_config dicts that accept any OpenAI-compatible endpoint. Ferro Labs AI Gateway is a drop-in.

Install​

pip install pyautogen

Configure​

import os
from autogen import ConversableAgent

config_list = [
    {
        "model": "claude-3-5-sonnet-20241022",      # routed to Anthropic by Ferro
        "api_key": os.environ["FERRO_API_KEY"],
        "base_url": os.environ["FERRO_BASE_URL"] + "/v1",
    }
]

agent = ConversableAgent(
    name="assistant",
    llm_config={"config_list": config_list},
)

reply = agent.generate_reply(messages=[{"role": "user", "content": "Hello from AutoGen"}])
print(reply)

To give different agents different providers, build per-agent config_lists with different model values.

Verify​

curl http://localhost:8080/v1/chat/completions \
  -H "Authorization: Bearer $FERRO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"gpt-4o","messages":[{"role":"user","content":"hi"}]}'

See also​

  • CrewAI — alternative multi-agent framework
  • LangGraph — explicit graph-based agents