Tue, Oct 27, 2026Los AngelesNew YorkChicagoHoustonPhoenixPhiladelphiaSan FranciscoSeattleAtlantaBoston
Herald Authority · 2950 North Loop West, Suite 500, Houston, TX 77092
Tech · Tech

Small Team Claims AI-Driven Model Development

Recursive, a company led by former Salesforce chief scientist Richard Socher, says its Eureka Machine automated the creation of its new model, Flint.

By Gabriel Fedorov ·

Illustration for: Small Team Claims AI-Driven Model Development
Illustration generated for this article, not a news photograph.

Recursive, a startup founded by former Salesforce chief scientist Richard Socher, claims its new model, Flint, was built largely through AI automation. The company's Eureka Machine reportedly handled the majority of the model's development pipeline.

The Eureka Machine created new training methods and model architectures, and generated code optimized for specific silicon chips. Recursive states that its automated system enabled a small team of researchers to build the model in a few months. The company shared benchmark scores indicating Flint outperforms similar-sized models on coding tasks.

Recursive raised $650 million at a $4.65 billion valuation from investors including Google Ventures, Greycroft, Nvidia, and AMD Ventures. Socher's cofounders include Jeff Clune, a professor at the University of British Columbia who researches indefinitely learning AI systems.

Recursive's approach relies on a system incapable of escaping human control, according to Socher. The Eureka Machine is intended to achieve full recursive self-improvement, but the process would still be overseen by humans who could stop it if it goes awry. Recursive views fears about self-improvement as unnecessary impediments to progress.

Some experts warn that letting AI improve itself could lead to disaster. A researcher who quit Anthropic this September warned that advances in AI, including recursive self-improvement, could lead to human extinction within a few years. Several bills in Congress propose to ban the technique.

It is unclear how much progress other US labs have made in this area. A team from Microsoft, Carnegie Mellon University, and the University of Virginia recently published work showing how an AI coding agent can recursively improve itself by modifying its own model's weights. Recursive has raised $650 million at a $4.65 billion valuation from Google Ventures, Greycroft, Nvidia, and AMD Ventures.

Socher says the approach used to build Flint could make cutting-edge AI more accessible. "It's the only way this very small team can produce a model that's actually better than any other model in its class," he told me. Recursive has raised $650 million at a $4.65 billion valuation from Google Ventures, Greyco.. [truncated]

Source: Wired · original report published

More Tech · Report a business

Get Herald Authority in your inboxThe stories that matter, every day. Free, a few times a week, unsubscribe any time.

Get Herald Authority in your inboxThe stories that matter, every day. Free, a few times a week, unsubscribe any time.