131 tokens a second: Recursive's Flint coding model runs on a laptop

By the GetNews desk · Edited by Sam Calder

Recursive released its first model, Flint, which is free, openly licensed and works offline on a laptop.

40Bmodel size; uses about 2.5B per task

Hype check

Hype6

Real5

As advertised

The noise matches the real change.

Our editorial opinion, based on the sources below. How we score

Your call: hype or real?

Real = it will matter in daily use. Hype = louder than it is. One vote, and it can’t be changed.

Our view and readers’ views are opinions, not statements of fact about a product.

Why it matters

A coding model that runs on your own machine keeps your code with you and needs no cloud subscription.

For most people

Skip: aimed at programmers.

Flint is a 40-billion-parameter mixture-of-experts model that uses about 2.5 billion per task.
It is free to download on Hugging Face under the Apache 2.0 licence.
In company tests it reached about 131 per second on a MacBook Pro with an M4 Pro chip.
The company says it beat other models under 50 billion on SWE-Bench Verified.
The company advises splitting big tasks into smaller steps, as it needs more guidance than large paid assistants.

Try itFree and open; the performance numbers come from the company itself.

One trusted outlet reports it. Not confirmed yet. Open the sources and check us.

Written with the help of AI tools and checked against the sources above. How we work

See a mistake? Tell us and we’ll fix it.