Mirendil has raised $1 billion in a Series A round led by Kleiner Perkins and Clearlake Capital. The San Francisco lab builds AI systems that carry out AI research and development, and it says the money will put frontier research within reach of smaller teams.
Battery Ventures, Doerr Capital, Gil & Co, Google, M13, Pear VC and S32 also took part, alongside other new and existing investors. Mirendil did not disclose a valuation. The round brings its total funding to $1.2 billion, after a $200 million seed round in June.
One researcher, one model, under $10 million
Mirendil says one researcher working with AI agents pretrained Mirendil-Base, a coding model, from scratch in three and a half months with less than $10 million of compute.
Agents wrote over 99 percent of the code, ran the experiments and diagnosed failures. The researcher decided what to try and what to trust.
The model sits in the 100 billion to 500 billion parameter range. Mirendil then post-trained it into Fern, its first model, which holds up to 10 million tokens of context, enough for days of an agent’s work at once.
Lor, the platform behind the lab
Mirendil also runs Lor, a platform where hundreds of agents write code and run experiments for days at a time. Lor is now in private beta.
The company says it runs on Lor every day. The platform is how it plans to bring the same way of working to other teams.
Built by former Anthropic researchers
Mirendil launched in June with a founding team of 20 researchers and engineers from Anthropic, xAI, Google DeepMind and OpenAI.
Chief executive Behnam Neyshabur co-led the science team at Anthropic and spent five years at Google DeepMind. Chief technology officer Harsh Mehta built the first version of Anthropic’s internal AI research platform.
The seed round in June was led by Andreessen Horowitz and Kleiner Perkins, with Nvidia among the investors. It valued the company at about $1 billion. The new round, closed five months later, is five times that size.
What the money will fund
The funding will advance the model work and broaden access to frontier AI development, the company said. Mirendil has also signed a multi-year partnership with Google Cloud, announced in August, to run training and inference on Google’s AI Hypercomputer.
“What excites us most isn’t just what Mirendil can accomplish as an AI lab. It’s what this approach could unlock for everyone else.”
Mamoon Hamid, partner at Kleiner Perkins
Mirendil says one person will be able to take on work that used to need a team, and a team will be able to take on work that used to need an institution. You can follow this year’s AI funding rounds on our funding tracker.
