Valued at $4.65 Billion, Recursive Unveils Flint, a 40-Billion Parameter Coding Model

Recursive, a startup focused on using artificial intelligence to build better artificial intelligence, has launched Flint, its first major open-weight model. The company is betting heavily on recursive self-improvement, a controversial method where AI systems are used to refine and upgrade their own development pipelines. This new release marks a significant step in the industry's shift toward autonomous model creation.

The financial backing behind this venture is substantial. Recursive has secured $650 million in funding, valuing the company at $4.65 billion. Investors include Google Ventures, Greycroft, Nvidia, and AMD Ventures. The Flint model itself contains 40 billion parameters, a size that allows it to run on high-end consumer hardware like a Mac with 48 GB of RAM, unlike the largest industry models that require industrial server racks.

The project is led by Richard Socher, formerly the chief scientist at Salesforce. He co-founded Recursive with Jeff Clune, a professor at the University of British Columbia who studies systems capable of indefinite learning. Socher states that the company utilized a recursive AI system called the Eureka Machine to automate most of the model's creation. This automation allowed a small research team to produce a world-class model in a matter of months.

Socher explains that the Eureka Machine handled nearly every aspect of the pipeline, from training the base model to optimizing code for specific silicon chips. The resulting model demonstrates strong performance in coding tasks. Benchmarks released by Recursive indicate that Flint outperforms other models with similar size and architecture. Socher notes that this efficiency is the only way such a small team could produce a model superior to its class.

The concept of recursive self-improvement remains a subject of intense debate. Some experts fear that allowing AI to improve itself indefinitely could lead to a loss of human control, potentially resulting in catastrophic outcomes. These concerns gained prominence this September when a researcher from Anthropic warned of potential human extinction within a few years. In response, several bills in Congress have proposed banning the practice. However, Recursive views these fears as overly dramatic and sees legislation as an unnecessary hurdle.

Socher emphasizes that the system used to build Flint is incapable of escaping human oversight. The goal is for the Eureka Machine to achieve full recursive self-improvement under human supervision, with the ability to stop the process if it goes wrong. He hopes this demonstrates that the necessary steps are in place to build such systems safely.

While other US labs continue to research this area, recent published work from Microsoft, Carnegie Mellon University, and the University of Virginia suggests that fully recursive systems are already feasible. For now, the approach appears to be a promising method for improving AI models and making cutting-edge technology more accessible to smaller teams.