Mirendil, an artificial-intelligence startup founded less than a year ago by former Anthropic researchers, raised $1 billion in a Series A round led by Kleiner Perkins and Clearlake Capital.
Battery Ventures, Doerr Capital, Gil & Co, Google, M13, Pear VC and S32 also took part, along with other new and existing investors. The company didn’t disclose its valuation. The round comes five months after a $200 million seed financing in June led by Andreessen Horowitz and Kleiner, which makes Kleiner the only firm to lead both rounds.
One researcher, many agents
Mirendil’s pitch is that AI agents can do most of the hands-on work of AI research, so a single person can take on projects that once needed a lab. The company says one researcher directing agents pretrained its Mirendil-Base coding model from scratch in about three and a half months for less than $10 million in computing costs, with agents writing more than 99% of the code. It then adapted the model into Fern, which can hold up to 10 million tokens of context, enough to review days of an agent’s work.
Its main product, Lor, is a cloud platform in private beta where hundreds of agents can write code, run experiments and provision computing power for days or weeks at a time. Mirendil says its own staff uses it daily. The company signed a multiyear Google Cloud deal in August for training and inference on Google’s TPU chips and Nvidia hardware.
Founders from the big labs
Mirendil was started in late 2025 by Behnam Neyshabur, the chief executive, and Harsh Mehta, the chief technology officer, both formerly of Anthropic, along with researchers who had worked at xAI and OpenAI. It remains a small team.
The financing adds to a busy month for AI fundraising, alongside OpenAI’s private round, Mercor’s $750 million raise and Anthropic’s initial public offering, which is set to price next week.
