Free models

Call real models. Pay nothing.

Models badged Free* on the Inpherio marketplace are free within limits, from any account with a verified email address. No card, no credit, no trial clock — the same API and SDKs as every paid model.

Free right now

Find them on the marketplace.

Every free model is badged Free* on the marketplace, with its own limits shown on its card.

See the free models on the marketplace

Which models are free, and their limits, can change.

How to use them

Exactly like a paid model.

  1. Create an account

    Sign up and verify your email address — free use needs a verified email.

  2. Make an API key

    In your dashboard, under API keys. No card or credit needed.

  3. Call a free model

    Use its model ID with any OpenAI-compatible SDK. Every response's X-Inpherio-Free-Tier header says free, or paid; reason=… with the limit that applied.

python · openai sdk
from openai import OpenAI
client = OpenAI(base_url="https://inpherio.co.uk/v1",
                api_key=INPHERIO_API_KEY)
reply = client.chat.completions.create(
    model="qwen3.5:2b",   # any model marked Free*
    messages=[{"role": "user", "content": "Hello, grid"}],
)

The limits, in full

What happens at a limit: nothing breaks.

  • Per model, per hour: each free model has its own hourly call limit for your account (shown above), and text models cap the reply length and prompt size of a free call.
  • Per day: up to 100 free calls a day across all free models, plus a daily token allowance; one free request at a time.
  • Past a limit the call is no longer free: with credit, it's billed at the model's listed price like any paid call; without credit, it's refused with a message giving that price. Never charged by surprise.
  • Content-checked: every free request is screened by a safety classifier first. Several blocked requests in a day pause free use for that account for 24 hours.
  • Paid traffic comes first: free requests use capacity paying customers aren't using, and a free stream may be stopped if a paid request needs the hardware.
  • No service level for free use — it can change or end at any time. The full rules are in the free usage terms.

When you need more, it's the same key.

Every model on the grid works with the key you already have — top up and pay per token, at each host's price.