Recursive, founded last year and led by former Salesforce chief scientist Richard Socher, emerged from stealth in May with $650 million in funding from backers including GV, Nvidia and AMD Ventures. The company's first model, Flint, is a public test of recursive self-improvement: the idea that AI can build better AI that in turn builds even better AI. This concept is debated in the industry, with some researchers concerned about AI slipping out of human control.
Flint is a mixture-of-experts model with 40 billion parameters, using only 2.5 billion at a time for any given task, making it light enough for an ordinary laptop. It is available for free on Hugging Face under the Apache 2.0 open-source license, allowing developers to use it offline, keep code private and avoid cloud usage fees.
In Recursive's tests, Flint generated about 131 tokens per second on a MacBook Pro with Apple's M4 Pro chip, roughly double the speed of similar models including Google's Gemma 4 and Nvidia's Nemotron 3.5 Lightning. On Nvidia's RTX Spark it topped 150 tokens per second. On SWE-Bench Verified, a benchmark for fixing real-world software bugs, Flint outperformed other models under 50 billion parameters. On Recursive's own average across six tests, only one model, Alibaba's Qwen3.8 27B, scored higher.
The company acknowledged Flint would need more guidance than large paid coding assistants and suggested breaking big tasks into smaller steps. Early reviews were positive. Wired's Will Knight used Flint on a Mac with 48GB of memory to build several small programs and found it performed well compared to alternatives.
Recursive built Flint using its automated research system, the Eureka Machine, which contributes ideas, runs experiments and uses results to decide next steps. The company said the system contributed to nearly every part of development, from pre-training to post-training and inference. Contributions include a sparser model shape, low-bit training that shrank Flint by 71% with almost no loss in quality, and custom kernels that speed up computations.
The longer-term goal is for the Eureka Machine to achieve full recursive self-improvement, with humans overseeing the process and able to stop it if something goes wrong. Recursive previously told TechCrunch that its $410 million compute deal with Amazon Web Services in June was about "agent count" rather than headcount, and said each new model should make the Eureka Machine a better researcher, building the next one faster.