Mirendil $1 Billion Series A Explained: Kleiner Perkins and Clearlake Back the AI Lab Teaching Agents to Do Research

Illustrative image: an artificial neural network with a chip — Mirendil has no public product screenshot; its Lor platform is in private beta (Wikimedia Commons, CC BY-SA).

TL;DR — Mirendil $1 billion Series A explained: the San Francisco AI lab founded by former Anthropic researchers raised $1 billion on 12 November 2026, led by Kleiner Perkins and Clearlake Capital, to build AI agents that do the hands-on work of AI research. The claim behind the money: one researcher working with agents pretrained a coding model from scratch in 3.5 months for under $10 million in compute.

Key Highlights

• $1 billion Series A announced 12 November 2026, led by Kleiner Perkins and Clearlake Capital, with Google, Battery Ventures, Doerr Capital, Gil & Co, M13, Pear VC and S32 also taking part.
• No valuation was disclosed. Bloomberg reported in September that Mirendil was in talks to raise at a $5 billion valuation — that figure is unconfirmed, and the announcement does not repeat it.
• The pitch: one researcher working with AI agents pretrained Mirendil-Base, a coding model in the 100–500 billion parameter range, from scratch in 3.5 months for less than $10 million in compute.
• Fern, the post-trained version of Mirendil-Base, holds up to 10 million tokens of context; Lor, the agent platform built on the same workflow, is in invitation-only private beta.
• Seed round was $200 million in June 2026 at a reported valuation of about $1 billion — total disclosed funding is now roughly $1.2 billion.

Mirendil $1 billion Series A explained: what the money is for

Mirendil is an AI research lab that wants to make frontier AI research something a small team — eventually one person — can do: "when agents do the hands-on work, the size of your team stops limiting what you can build." The $1 billion is meant to turn that idea from a single experiment into a product other scientists and engineers can use.

The company was founded in early 2026 by chief executive Behnam Neyshabur and chief technology officer Harsh Mehta, both former Anthropic researchers who left in December 2025; the 20-person founding team came from Anthropic, xAI, Google DeepMind and OpenAI. Mirendil emerged from stealth on 24 June 2026 with a $200 million seed round led by Andreessen Horowitz and Kleiner Perkins at a reported valuation of about $1 billion, with Nvidia investing afterwards. Google is both an investor and a supplier, through a multi-year Google Cloud partnership announced in August.

The round is the largest US funding round of the week of 7–13 November, according to Crunchbase, ahead of Mercor's $750 million — and it caps a strong week for agent infrastructure. LangChain raised $300 million at a $3 billion valuation, and Fractile raised $720 million at a $7 billion valuation in the same window. Investors are paying for the plumbing that makes AI agents usable, not just for bigger models.

The claim behind the round: one researcher, 3.5 months, under $10 million

Three weeks before the raise, Mirendil published the experiment the round rests on (21 October 2026). One researcher, directing AI agents, pretrained Mirendil-Base — a 100–500 billion parameter coding model — in 3.5 months for less than $10 million in compute; the company says agents wrote more than 99% of the pretraining code. It then post-trained the model into Fern 0.1, with up to 10 million tokens of context.

Lor is the product that turns that internal workflow into something others can use. Introduced on 28 October 2026, it is a cloud-based system whose agents write code, run experiments, provision compute and operate on machines the user selects — "hundreds of agents writing code and running experiments for days at a time", according to the announcement. It is in invitation-only private beta, free during the beta.

Detail	What is confirmed
Amount	$1 billion, Series A, announced 12 November 2026
Lead investors	Kleiner Perkins and Clearlake Capital
Other participants	Battery Ventures, Doerr Capital, Gil & Co, Google, M13, Pear VC, S32
Valuation	Not disclosed; Bloomberg reported talks at $5 billion in September (unconfirmed)
Total raised to date	About $1.2 billion, including the $200 million seed of June 2026

The honest caveats matter as much as the headline. None of the performance numbers has been independently checked: there is no outside evaluation of Mirendil-Base or Fern, and no public benchmark suite behind the "competitive coding model" description. The company's own published comparison put Mirendil-Base behind Kimi K3 on bits-per-byte (0.222 against 0.193, where lower is better). A billion dollars does not verify the claim; it finances the attempt to prove it at scale.

Illustrative image: a brain under artificial intelligence control (Wikimedia Commons, CC0, via Pixabay).

Why it matters for readers in India

There is no India-specific data in the announcement — no Indian investors, customers or hiring plans were named. But the audience for what Mirendil is selling is exactly where India is strong: developers and small research teams. If agents can genuinely take over the hands-on engineering of AI research, the advantage shifts from whoever can hire the biggest lab to whoever can direct agents well — and India has one of the world's largest developer bases and a fast-growing crop of small AI startups that cannot afford a frontier lab.

The timing is also notable. Dealroom points out that the round lands a week before Anthropic — the lab both founders left — is set to list on Nasdaq. Indian readers watching the AI race should watch Lor's beta, not the valuation: an invitation-only tool that works for real researchers is the proof, and it is the one thing money alone cannot buy.

Should you care?

If you build with AI agents, yes — watch Lor's private beta, not the $1 billion. The round is real and the investors are serious, but the core claim is still company-reported and untested by outsiders. If you are a founder or investor, the signal is the pattern: agent infrastructure took the three biggest cheques of the week. If you only follow AI news casually, the takeaway is simple — the race has moved from building bigger models to building agents that build models.

FAQs

What is Mirendil?
Mirendil is a San Francisco AI research lab founded in early 2026 by former Anthropic researchers Behnam Neyshabur and Harsh Mehta. It builds AI agents that do the hands-on work of AI research — writing code, running experiments and provisioning compute — so a small team or a single researcher can attempt projects that used to need an institution.

How much did Mirendil raise, and who led the round?
Mirendil raised $1 billion in a Series A announced on 12 November 2026, led by Kleiner Perkins and Clearlake Capital. Battery Ventures, Doerr Capital, Gil & Co, Google, M13, Pear VC and S32 also participated, following a $200 million seed round in June 2026 — total disclosed funding is about $1.2 billion.

What is Mirendil's valuation?
Mirendil did not disclose a valuation for the Series A. Bloomberg reported in September 2026 that the company was in talks to raise at a $5 billion valuation, but no source has confirmed the final terms match that report. The June 2026 seed was reported at a valuation of about $1 billion.

What is Lor, and what is Mirendil-Base?
Mirendil-Base is the coding model Mirendil says one researcher pretrained from scratch in 3.5 months with AI agents, for less than $10 million in compute. Fern is its post-trained successor, with up to 10 million tokens of context. Lor is the cloud platform that turns that workflow into a product — agents that write code and run experiments for days — now in invitation-only private beta.

Takeaway

A $1 billion Series A is a bet that AI agents can do the work of an entire research lab — and the only evidence so far is one company-reported experiment. The valuation is undisclosed, the benchmarks are unverified, and the product is in private beta. Watch what Lor does in real researchers' hands; that is what this money is buying.

If you build with AI agents, the related read is our LangChain $300 million funding explainer — the other big agent-infrastructure round of the same week.

Sources

Mirendil — Mirendil Raises $1B to Democratize Frontier AI R&D (official announcement, 12 November 2026)
PR Newswire — Mirendil Raises $1B to Democratize Frontier AI R&D (12 November 2026)
Mirendil — Pretraining at Scale Without a Pretraining Team (research report, 21 October 2026)
Crunchbase News — The Week's 10 Biggest Funding Rounds: Agentic AI and Infrastructure Are Still on a Roll (13 November 2026)
Dealroom — Ex-Anthropic Researchers' Mirendil Raises $1B Series A Led by Kleiner Perkins and Clearlake (12 November 2026)
Tech Times — AI Lab Mirendil Raises $1B Series A: Release Names No Valuation, Cites Only Company Tests (13 November 2026)