Deep Cogito raises $43M to help companies own their AI models

What's the deal? Deep Cogito, a San Francisco AI research lab, raised $43 million in a Series A round to develop specialised AI models that companies can own. TQ VenturesDealroom has a profile for this one. Try Dealroom → led the round, with participation from Benchmark, Nexus Venture Partners, Atreides ManagementDealroom has a profile for this one. Try Dealroom →, South Park Commons, and cybersecurity firm Zscaler. The raise brings the startup's total funding to more than $56 million.

What's the endgame? Deep Cogito builds open-weight models — which publicly share their underlying numerical parameters and are cheaper to run than proprietary ones. Its post-training engine powers its Cogito family of frontier models and specialised models trained on an enterprise's proprietary data.

Why now? Founded in 2024 by two former GoogleDealroom has a profile for this one. Try Dealroom → employees, the company is betting more firms will build their own models with outside help. "The idea that you are going to use open models…for your own proprietary intelligence, and optimize it for your particular content has resonated with a lot of folks," chief executive Drishan Arora said. "I believe there's an inflection point here."

How it works: The lab focuses on post-training, the phase after a model has learned from general data. It applies large-scale reinforcement learning and recursive self-improvement, in which a model teaches itself to improve for compounding gains. The long-term goal is models that "progressively improve their own capabilities and ultimately move beyond the limits of human-generated training data."

Why it matters: Cost is one draw, but not the only one, according to Arora. "Specialized models trained on product-specific data often outperform general frontier models on that product," he said.

The signal: The $43 million round sits in the top 10% of US Series A deals over the past four years, a sign of investor appetite for open-weight infrastructure. China dominates the market for such models, and backers see room for a homegrown alternative. "We think of them becoming one of the defining companies building open-weight models and specialized enterprise intelligence," said TQ Ventures co-founding partner Schuster Tanger. "We think there's an aperture for a highly proficient American solution."

Read more: The Wall Street Journal

Image credit: Deep Cogito