Recursive Unveils Flint, a 40-Billion Parameter Coding Model Built by AI

A new open-weight AI coding model named Flint has arrived, marking a significant step toward the controversial concept of recursive self-improvement, or RSI. The model, developed by the startup Recursive, suggests that using artificial intelligence to build and refine other artificial intelligence systems is becoming a practical reality rather than just a theoretical goal. For the Southeast tech community and broader industry observers, this development signals a shift in how high-performance models are created, potentially lowering the barrier to entry for advanced AI capabilities.

The path to Flint’s release began with the establishment of Recursive, a company founded with the explicit goal of leveraging RSI to advance AI and other scientific fields. Richard Socher, the company’s CEO and co-founder, explains that nearly every aspect of the model’s development pipeline was handled and improved by AI. The team utilized an internal system called the Eureka Machine to automate the majority of the development process. This automation allowed a small group of researchers to construct a world-class model in a matter of months, a timeline that would typically be much longer for a team of that size.

The Eureka Machine did not just assist; it devised specific innovations such as new training methods for the base model, novel architecture designs, and code optimized for specific silicon chips. Recursive has shared benchmark scores indicating that Flint outperforms other models of similar size and architecture on coding tasks. This is notable because AI is already widely used to help build and test other AI models, but creating a loop where the AI decides on its own improvements and initiates another round of development represents a more aggressive approach to the process.

The concept of RSI has sparked significant debate within the tech industry. Some experts worry that allowing AI to improve itself indefinitely could lead to unintended consequences, with fears that such systems might eventually escape human control. These concerns gained traction earlier this year when a researcher left Anthropic to warn that rapid advances in AI, including RSI, could pose existential risks. The issue has become so prominent that several bills in Congress have proposed legislation to ban recursive self-improvement, although Recursive views these fears as more science fiction than immediate reality.

Socher emphasizes that Flint was built using a system designed to remain under human oversight. While the goal is for the Eureka Machine to eventually achieve full recursive self-improvement, humans will still be in the loop to stop the process if necessary. The company has secured $650 million in funding at a $4.65 billion valuation from investors including Google Ventures, Greycroft, Nvidia, and AMD Ventures. The team includes Jeff Clune, a professor at the University of British Columbia who researches AI systems capable of indefinite learning, alongside Socher, who previously served as chief scientist at Salesforce.

In practical testing, Flint has shown impressive capabilities for its size. With 40 billion parameters, it runs on modest hardware, such as a Mac with 48 GB of RAM, unlike massive models that require industrial server racks. Users have reported that Flint is fast and highly capable, successfully building command-line tools, web scrapers, and explainer websites without the errors or slowness often associated with models of this scale. While larger models exist, Flint’s performance suggests that recursive development can yield highly effective results even from smaller teams.

Other research institutions are also exploring similar paths, with recent work from Microsoft, Carnegie Mellon University, and the University of Virginia demonstrating how AI coding agents can recursively improve their own software harnesses and weights. This convergence of efforts suggests that fully recursive systems are becoming feasible. For now, the technology appears to be a promising method for enhancing AI models, making cutting-edge technology more accessible to a wider range of developers and organizations.

As the industry moves forward, the focus will remain on balancing the efficiency gains of RSI with the need for robust safety measures. Recursive believes that the approach used to build Flint can make advanced AI more attainable, proving that small teams can produce models that rival much larger efforts. This development may reshape expectations for what is possible in AI development, particularly for organizations that lack the massive computational resources of the industry giants.