They say the model achieves performance comparable to Grok 4 proper with 40% less token use, and even pushes the price of those tokens down by 98%.
This makes it one of the cheapest and most efficient models out there, xAI claims, but many say they still prefer GPT-5 mini, which is also plenty cheap.
The model has a 2 million token window, and can switch between reasoning and non-reasoning on the fly.
Travels well in benchmarks
In benchmarks posted by xAI, it appears to score slightly below the grown up Grok 4 model and is a little behind GPT 5-high.
On LMarena it reaches joint eight place on text tasks, and is just ahead in first place on search tasks, but is nowhere to be seen on the other leaderboards.
Available for all
It’s the cost per token that is the showstopper here, for a top notch reasoning model — at $0.20 per input and $0.50 for output, its cost-efficiency is unmatched, xAI claims.
It is so cheap and competent, xAI say, that they are making it available for free in the app on the web, so you can check it out yourselves.
— For the first time, all users, including free users, will have access to our latest model without restrictions, marking a step toward democratizing advanced AI, xAI writes.
Read more: xAI’s announcement, writeup by Engadget, discussion on r/Singularity.