Mirendil raises $1 billion Series A to let AI agents do frontier AI research

San Francisco-based startup Mirendil has raised $1 billion in a Series A round to make frontier AI research and development accessible to more scientists and engineers.

The investors

Kleiner Perkins and Clearlake Capital co-led the round, joined by Battery Ventures, Doerr Capital, Gil & Co, Google, M13, Pear VC and S32, among other new and existing investors. The company did not disclose its valuation. The round is unusually large for an early-stage company.

According to SiliconANGLE, Mirendil was launched last year by two former Google DeepMind and Anthropic researchers. CEO Behnam Neyshabur is a co-inventor of SAM, a popular algorithm that helps models perform better on tasks they were not specifically trained for, while CTO Harsh Mehta worked on pretraining projects.

One researcher, hundreds of agents

Mirendil's main product, Lor, is a platform in private beta that runs hundreds of AI agents writing code and running experiments for days at a time. The company says it uses Lor every day for its own work.

Mirendil says one researcher working with agents pretrained Mirendil-Base, a competitive coding model, from scratch in 3.5 months with less than $10 million in compute. SiliconANGLE reports that agents wrote more than 99% of the model's code. The company then post-trained the model into Fern, which can hold up to 10 million tokens of context.

"When agents do the hands-on work, the size of your team stops limiting what you can build," the company said. Mirendil plans to use the money to advance its agent-centric approach, widen access to Lor and hire.

Sources: Mirendil, SiliconANGLE, Silicon Report