Mirendil has raised a $1 billion Series A while its AI research platform remains in private beta, the company announced on November 12. Kleiner Perkins and Clearlake Capital co-led the round.
Mirendil targets scientists and engineers who want to train and improve their own AI models. The company says Lor lets researchers assign batches of experiments to agents that arrange GPU computing and keep jobs running. It supports models from other labs alongside Mirendil’s own, with access invitation-only and free during the beta.
The new round is five times the $200 million seed financing Mirendil announced on June 24, co-led by Andreessen Horowitz and Kleiner Perkins. Google and Battery Ventures are among the Series A participants. Mirendil did not disclose a valuation for the new round.
In an October 21 research post, Mirendil said one researcher working with agents built its Mirendil-Base coding model in three and a half months using less than $10 million in compute. That figure included exploratory experiments, the final training run, agent usage and data processing.
The researcher directed the project and assessed its results. Mirendil credits the agents with writing more than 99% of the code.