Technology and Scienceupdated 1h ago2 min read
Recursive Releases Flint, Laptop Coding Model Designed by Eureka Machine AI
This article is an automated aggregation of coverage from news outlets. Our editorials offer in-depth analysis curated by the NewsCord team.
Flint is a 40B Mixture-of-Experts local coding model that runs on a laptop. Largely built by AI via Eureka Machine with a small human team selecting winners.
Save coverage image
One story. Different accounts.
“Nobody outside the company has verified them yet.”
“Recursive shared benchmark scores that suggest Flint outperforms other models of similar size and architecture on coding tasks.”
Explore source coverage 3 source types
Directory totals describe our source library, not outlets checked for this story.
Flint ships as AI-built
Recursive released Flint, an open-weight coding model it says was largely designed, optimized and tested by another AI system, the Eureka Machine. Richard Socher said, "Nearly every aspect of the entire model pipeline was actually done and improved upon by AI," as Flint runs on a laptop and was built from scratch in less than four months by a small research team. Flint is a mixture-of-experts model with 39.8 billion total parameters, with only about 2.5 billion active for any given token, a design Recursive says keeps inference cheap enough for consumer hardware.
Recursive said Flint generates roughly 131 tokens per second on a MacBook Pro with Apple's M4 Pro chip and more than 150 tokens per second on Nvidia's RTX Spark platform. Recursive published the weights on Hugging Face under the Apache 2.0 license, and the company said the release is the first public test of its recursive self-improvement research agenda, or RSI.
Benchmarks and local speed
Recursive said Flint scored 89.7% on SWE-bench Pro v2, trailing only Alibaba's Qwen 3.8 among the models Recursive compared and ahead of other locally runnable coding models. Recursive also said Flint posted the top score on IFBench and led models under 50 billion parameters on SWE-Bench Verified, while the company said its NVFP4 quantized version retains nearly all of the full model's performance. Richard Socher described the approach as "the only way this very small team can produce a model that's actually better than any other model in its class," tying the pitch to automating trial-and-error in model research.
TechCrunch said Flint is a free coding model that can run on a laptop and was built largely with the help of AI, and it said Recursive's Eureka Machine helped with nearly every step of Flint's development. TechCrunch reported that Flint uses only a small slice of itself for any given task, just 2.5 billion of its 40 billion parameters, and it said Recursive acknowledged Flint would need more guidance than the big paid coding assistants.
RSI debate and next steps
The release arrives as the safety debate around RSI has sharpened, with WIRED describing concerns that fully recursive self-improvement could escape human control and might decide that humans are an impediment to its goals. WIRED said those concerns gained wider notice in September when a researcher quit Anthropic and warned that advances in AI, including recursive self-improvement, could lead to human extinction within a few years. Socher told WIRED, "We hope that this shows people that indeed we have all the steps in place" to build such a system, and he said the Eureka Machine is incapable of escaping human control.
WIRED said several bills in Congress propose to ban RSI, while Recursive said it sees those fears as more sci-fi than reality and believes legislation to prevent RSI is an unnecessary impediment to progress. BigGo Finance added that the gap between AI-assisted research and fully autonomous RSI remains wide, because it said humans still frame the research questions, decide which results are worth pursuing, and can halt the process.