One endpoint for the free tier — routed by difficulty across your workspace's free models.
One endpoint for the free tier. Instead of picking a specific -free alias yourself, call orcarouter/free and it routes each request by difficulty across your workspace's free models — never touching your wallet, with free use capped by request rate rather than a fixed number of calls.
Each request is scored on prompt length, code density, tools and reasoning cues — the same signal our adaptive router uses.
Mundane prompts go to the light free tier; hard ones to the stronger free tier — drawn from your workspace’s free models.
Model usage costs nothing — both tiers are priced at $0, so it makes no difference which one serves the request.
Free access is limited by request rate, not by a balance — the model usage itself costs nothing, and the allowance never runs out.
Every workspace gets a request-rate allowance. It grows with what you have paid for over the account’s lifetime, so topping up raises the free limits too.
Individual free models can carry their own tighter rate, so a heavily-loaded one throttles without affecting the rest.
import os
from openai import OpenAI
client = OpenAI(
base_url="https://api.orcarouter.ai/v1",
api_key=os.environ["ORCAROUTER_API_KEY"],
)
response = client.chat.completions.create(
model="orcarouter/free",
messages=[{"role": "user", "content": "Hello"}],
)
print(response.choices[0].message.content)