Mirendil (@mirendil) header
Mirendil (@mirendil) profile picture

Democratizing frontier AI R&D to accelerate science and technology.

San Francisco
Joined March 2026
Photo by Mirendil (@mirendil)Photo by Mirendil (@mirendil)Photo by Mirendil (@mirendil)Photo by Mirendil (@mirendil)Photo by Mirendil (@mirendil)Photo by Mirendil (@mirendil)Photo by Mirendil (@mirendil)Photo by Mirendil (@mirendil)Photo by Mirendil (@mirendil)Photo by Mirendil (@mirendil)
Today we're announcing our $1B Series A, led by @kleinerperkins and @clearlake. We believe making frontier AI R&D more accessible is the most direct path to AI that solves humanity's most pressing problems across science and technology.
50
61
57
744
223,066
One researcher, working with agents, pretrained Mirendil-Base from scratch in 3.5 months. We post-trained it into Fern, which holds up to 10M tokens of context. Lor, now in private beta, runs hundreds of agents for days at a time.
1
1
3
29
6,856
Huge thanks to @baseten! Fast, reliable rollouts and seamless weight syncs kept our RL run moving at scale. A great partner across the board.
.@mirendil is building self-accelerating AI R&D to solve humanity's most pressing problems. To scale RL training across a global fleet of thousands of GPUs, it runs frontier RL on Baseten. We power their rollout fleet with new policies served in ~40 seconds, and sampling optimized for 1.9× higher throughput. Learn more about our collaboration here: baseten.co/resources/custome…
4
11
1
81
9,356
Meet Lor. We’re building Lor to do the work of an entire AI lab. Today, its agents write our code, run experiments, and iterate on the results for days, on any machine, hundreds at a time. Now in private beta ↓
62
82
27
876
136,757
Proteins, materials, medical scans: more and more of science runs on AI. Building it takes a team. It doesn’t have to. One Mirendil researcher just pretrained Mirendil-Base in 3.5 months: bit.ly/4ssSbEV We are building Lor to be a frontier AI lab for every scientist.
1
35
3,418
Lor is in private beta: free during the beta, on macOS, Linux, and web, running Claude, GPT-6, Gemini, open models, and our own Fern. We’d love to work with the people who’ll push it the hardest. Join the waitlist: mirendil.com/lor-beta/
3
46
3,194
What you can hand off to Lor: “Run these 12 ablations overnight and write a report.” “Audit all files in this repo for security bugs & open a PR for each fix.” “Reproduce Table 2 of this paper on 8 GPUs & explain where our numbers differ.” Read more: bit.ly/4kl5e9l
1
2
212
Run hundreds of agents at once. Much of building AI is parallel: reading thousands of agent traces, running hundreds of experiments, building training environments. In one run, 102 agents in Lor built environments from 100 PyTorch PRs, averaging 8 minutes each.
2
170
Introducing Fern, Mirendil’s first model: pretrained from scratch, with up to 10 million tokens of context. That’s ~0.5M lines of code or days of an agent’s work - all in context at once. Built for high-volume agentic work in AI R&D: auditing training data, debugging training runs from logs, and reading agent traces end to end. Fern’s next-token prediction keeps improving as context grows to 10M tokens. Fern 0.1 is Mirendil-Base after a short round of post-training. It’s the first and smallest of three models, with two larger ones now being pretrained. We’ll ship frequent updates and make the model available soon.
38
75
46
844
68,399
After a short round of post-training, here’s where Fern 0.1 stands on agentic benchmarks. This is our first version, with frequent updates ahead.
1
33
3,982
Mirendil retweeted
We ran @mirendil's new base model on Baseten's private codebases; that is, code it's never seen. Mirendil-Base-Large had the lowest bits per byte (a tokenizer-independent measure of perplexity) of any model we tested. We're excited to be the inference partner to serve Mirendil models. Congrats to the Mirendil team on this launch!
9
19
11
235
61,010