CrewAI
Multi-agent crews on an OpenAI-compatible LLM
CrewAI accepts an LLM pointed at any OpenAI-compatible /v1 host. Use
the openai/ model prefix.
| Base URL | https://api.orientx.ai/v1 |
| Auth | ORIENTX_API_KEY |
| Model | openai/<MODEL_ID> |
Install
pip install crewaiAttach the LLM
import os
from crewai import Agent, Crew, LLM, Task
llm = LLM(
model="openai/<MODEL_ID>",
base_url="https://api.orientx.ai/v1",
api_key=os.environ["ORIENTX_API_KEY"],
)
researcher = Agent(
role="Researcher",
goal="Summarize what an API gateway does",
backstory="You write short, precise technical notes.",
llm=llm,
)
task = Task(
description="Explain an API gateway in two sentences.",
expected_output="Two sentences.",
agent=researcher,
)
Crew(agents=[researcher], tasks=[task]).kickoff()Pass the same llm= into every agent that should call orientx.
You can also export OPENAI_API_BASE=https://api.orientx.ai/v1 and OPENAI_API_KEY instead of
passing base_url / api_key into LLM().