OptraCloudAI Inference · powered by Globussoft Start free — 150M tokens
Free tier · no card

150,000,000 tokens, on us.

Verify an email and a mobile number and start generating. No credit card, no trial clock, and no meter running while you think.

In units you actually think in

What 150,000,000 tokens buys.

A token is an abstraction nobody buys. Every divisor below is a real generation measured on this endpoint, so you can check it.

What 150,000,000 tokens actually buys

You could generate roughly this many
Complete web pages the demo page was 2,973 tokens, start to finish 50,454
Production functions, with edge cases a real one from our test suite, not a toy 619,834
Documents summarised reading costs nothing — you pay only for the answer 300,000
Chat sessions about ten turns each 50,000
READMEs or release notes four sections and a runnable sample 326,086

Divisors are real generations measured on this endpoint, not estimates. At 60 token/sec that is about 694 hours of continuous output — and since we bill only what the model writes, a fifty-page PDF costs nothing to read. Even used hard, all day, every day, most developers will not finish it in a year.

The same model, priced by other people

What it would cost you elsewhere.

What that would cost you elsewhere

Buy the same model from 150,000,000 output tokens
Chutes managed inference · $3.000 per million output tokens $450
AkashML managed inference · $3.200 per million output tokens $480
OptraCloud the same model, our own hardware in India Free

Published output-token prices for Qwen3.8-27B, the exact model we serve, verified against each provider’s own endpoint data in August 2026. Two rows, not more, because those are the only providers serving these weights so far — a model released this month is not on the big catalogues yet. It is the same weights, priced by people who also have to run them.

Bring someone with you

Another 150,000,000, to both of you.

Share your invite link from the portal. Once the person you invited has generated 100,000 tokens — about one afternoon of real work — you each receive another 150,000,000.

Why it pays late

On use, not signups

Paying on signup rewards whoever can create accounts fastest. Paying on 100,000 generated tokens rewards bringing someone who actually wanted this.

Why we ask for a number

One grant per person

A verified mobile is the only thing standing between this offer and somebody collecting it a hundred times. It is also the only personal detail we ask for.

When it runs out

Nothing breaks

Your key keeps working on a paid plan from ₹4,999 a month, or invite someone and carry on. No card on file means no surprise charge.

Free generation runs at 60 token/sec. Paid plans hold a 60 token/sec floor even when every seat is busy.