Recursive releases Flint, a free laptop coding model built largely by AI

Recursive, an AI startup founded last year, has released its first model, Flint. The company says its automated research system, which it calls the Eureka Machine, helped with nearly every step of development, letting a small team build the model from scratch in under four months.

Flint activates only 2.5 billion of its 40 billion parameters for any given task, which keeps it light enough for an ordinary laptop. In Recursive's tests on a MacBook Pro with Apple's M4 Pro chip, it generated about 131 tokens per second, roughly twice the speed of similar models it compared against. On SWE-Bench Verified, it outperformed other models under 50 billion parameters.

Recursive acknowledges that Flint needs more guidance than large paid coding assistants and suggests breaking big tasks into smaller steps. Wired's Will Knight found it handled his small coding requests reliably, where smaller models often ran into errors. Because it runs locally, developers can work offline and keep their code private without usage fees.

Developers building coding agents can run a capable model locally, avoiding per-token cloud charges and keeping code off remote servers. Flint's speed on consumer hardware makes local inference a practical option for agent loops. Recursive says the model needs more step-by-step guidance than paid assistants, so agents should split large tasks before handing them over.