LE PERSPECTIVE / THE STORY

Mirendil has raised $1 billion in a Series A to make frontier AI research possible with much smaller teams, betting that AI agents can handle much of the work behind developing advanced models.

Kleiner Perkins and Clearlake Capital led the round. Battery Ventures, Doerr Capital, Gil & Co, Google, M13, Pear VC and S32 participated alongside other investors.

The Research Team Gets Smaller
Mirendil says one researcher pretrained a coding model in 3.5 months for less than $10 million in compute. Agents wrote more than 99% of the code, but the researcher still chose what to test and what to trust.

A $1B Round, Months After Seed

Mirendil emerged publicly in June with a $200 million seed round led by Andreessen Horowitz and Kleiner Perkins, followed by investment from Nvidia. Its founding team brought together researchers and engineers from Anthropic, Google DeepMind, OpenAI and xAI.

The new funding supports a different kind of AI lab: one where researchers set direction while agents run experiments, debug failures and do repetitive engineering work.

Inside Mirendil’s AI Research Stack

Mirendil-Base, its first coding-focused model, was pretrained from scratch. The company says the experiment took one researcher three and a half months, with agents doing almost all the hands-on coding.

Its model Fern supports context windows of up to 10 million tokens. Lor, the company’s agent platform, can coordinate hundreds of agents, allocate computing resources and run experiments over multiple days. Lor remains in private beta.

The Test Ahead

Mirendil wants to extend this approach beyond software development to accelerate research in science and technology. But the results disclosed so far are company-reported, and they do not demonstrate a fully autonomous research lab.

The next question is whether the same model works at larger scales, when experiments become more costly and scientific judgment matters even more. This $1 billion round is a wager that better AI research may ultimately depend on redesigning the lab itself.

Get stories like this in your inbox every week. Subscribe to The Week in Perspective.