A better place to think.
Free to start. Open a chat and ask Wren the thing you've been chewing on.
Wren is a warm AI thinking partner. Ask, reason, and draft — and when it makes something, watch it render live in a panel beside the chat.
Wren is a chat you'll actually keep open — because it doesn't just answer, it shows the reasoning and hands you something you can use.
Watch Wren think with a live shimmer, then stream a clean, well-typeset answer. Stop, regenerate, or steer — you're never waiting on a black box.
Open a chat →Ask Wren to build a page, app, chart, or SVG and it opens in Frames — a live preview beside the chat, versioned and shareable.
Explore Frames →Group related chats and give Wren standing instructions — a tone, a brief, the facts it should always keep in mind for that body of work.
See Projects →Dial the voice — concise, explanatory, formal — and pick the model: Seed for speed, Field for balance, Grove for extended thinking.
Try the picker →The best part of Wren isn't the answer — it's the seam between thinking and making. Reasoning shimmers into view, prose streams in serif, and the thing you asked for renders itself in Frames while you read.
Named after how things grow. Pick for the task in front of you — every tier is served through the same hosted frontier model, tuned for its job. We say so plainly.
How they score →Fastest. Quick drafts, edits, and back-and-forth.
The everyday default — balanced quality and speed.
Most capable. Extended thinking for hard problems.
Every week Wren re-runs a small, scripted eval on itself and publishes the numbers — the good ones and the ones that flatter nobody. No third-party leaderboard, no cherry-picking. Just receipts we actually have.
“The Frames panel changed how I prototype. I describe a landing page, it renders, I say ‘make the hero warmer’ and I get v2. It's the fastest sketchpad I have.”
“I write grant proposals for a small nonprofit. Wren's Projects hold our voice and our facts so I stop re-explaining ourselves every session.”
“I trust a tool more when it shows the reasoning and admits limits. The public evals sold me before I ever signed up.”
“Grove's extended thinking is my study buddy for organic chem. It works the problem out loud, then I check my own steps against it.”
“Concise style + Seed model is my email drafting setup. Fast, tight, done. Then I switch to Field when I actually need to think.”
“It reads like a manuscript, not a chat app. The serif prose sounds like a person thought about it. I didn't expect to care, but I do.”
Wren is a web-based AI chat product — a thinking partner for writing, reasoning, and building. It's built on hosted frontier language models; we're a product company, not a model lab, and we're careful to say so. What we add is the craft: the streaming, the Frames panel, Projects, and honest, public evals.
When you ask Wren to build a web page, small app, chart, or SVG, it opens the result in Frames — a live preview panel beside your chat, with Code and Preview tabs, a version stepper, and one-click publish to a shareable link. Ask for a change and you get a new version, in place.
You choose a tier — Seed, Field, or Grove — tuned for speed, balance, or extended thinking. In this reference build every tier is served through the same hosted frontier model with different settings; a production Wren would route each tier to a distinct model. We publish exactly how they score on our evals page.
No. Your chats and Projects are yours; we don't train on them. See our privacy page and data-use notes for the specifics.
Free gives you real Wren: chat, Frames, Projects, and the Seed & Field models with fair daily usage. Pro ($17/mo) adds Grove, higher limits, and unlimited Projects & Frames. Max and Team scale usage and collaboration. Full breakdown →
Free to start. Open a chat and ask Wren the thing you've been chewing on.