Craft
The margin is the message
Why Wren shows its reasoning and renders in Frames instead of hiding the work. A design philosophy: the manuscript, not the messenger.
Named launches, craft essays, and honest research. We build a warm AI thinking partner on hosted frontier models — no lab mystique, no hype, and we show our work here the same way Wren does.
Our most capable tier, built for the problems worth slowing down for. Grove thinks longer, out loud, and hands the reasoning back to you. Here's what it's good at, where it isn't, and the eval numbers — flattering and not.
Read the launch →Why Wren shows its reasoning and renders in Frames instead of hiding the work. A design philosophy: the manuscript, not the messenger.
Inside Wren's weekly self-eval harness — why we publish the unflattering numbers, and how to actually read the evals page.
We're a product on hosted frontier models — not a model lab. Here's the philosophy behind Field Notes, public evals, and no hype.
The flagship extended-thinking tier, now on Pro. What it's for, honest eval numbers, and the limits we're still working on.