Recursive, an AI startup founded last year, is starting to live up to its name. The company, which aims to develop AI that can improve itself, released its first model, Flint, on Wednesday. Flint is a free coding model that can run on a laptop and was built largely with the help of AI. Recursive’s automated research system, known as the Eureka Machine, assisted in nearly every step of Flint’s development, allowing a small team to build the model from scratch in under four months.

The Concept of Recursive Self-Improvement

Flint serves as an early public test of recursive self-improvement: the idea that AI can build better AI, which in turn can create even better AI. This concept is one of the most debated in the industry, with some researchers expressing concerns that self-improving AI could slip out of human control. Currently, Recursive’s version is more limited, functioning as an AI research assistant overseen by humans, producing a model small enough for a laptop.

Funding and Leadership

The company emerged from stealth mode in May with $650 million in funding at a $4.65 billion valuation, backed by investors including GV, Nvidia, and AMD Ventures. Recursive is led by CEO Richard Socher, former chief scientist at Salesforce.

Features of Flint

Because Flint operates on a user’s own computer rather than on cloud servers, developers can use it offline, maintain the privacy of their code, and avoid the usage fees associated with AI cloud services based on text processing. The model is available for free download on Hugging Face, a popular platform for sharing AI models, under the Apache 2.0 license, which allows for free use and modification.

Flint is a mixture-of-experts model, utilizing only a small portion of its capabilities for any given task: just 2.5 billion of its 40 billion parameters at a time. This design makes it lightweight enough for an ordinary laptop and significantly smaller than Meta’s Muse Spark Open, which was released last week and requires 816 billion parameters and at least a terabyte of memory.