Real answers for getting the most out of Wren — chat, Frames, Projects, models, billing, and the honest details about your data.
No answers matched that. Try a different word, or send us a note.
Your first minutes with Wren.
Open the app and type into the message box at the bottom — that's it. Press Enter to send, Shift+Enter for a new line. You'll see Wren think with a live shimmer, then stream its answer in serif prose. From the homepage you can also type your question into the “Ask Wren anything” box, which carries it straight into a new chat.
You need a free account to save chats, but no card is required to start. The Free plan gives you real Wren — Chat, Frames, Projects, and the Seed & Field models — with fair daily usage. See the full breakdown on Pricing.
No. Wren is web-only — it runs in your browser on desktop or mobile, in a single workspace. There's no native iOS or Android app and no separate API product. We'd rather do one surface well than spread thin, and we say so plainly.
Yes. While Wren is responding, a Stop control appears — use it any time. After an answer lands you can Regenerate for a fresh take, or just reply to steer it. Nothing is a black box you have to wait out.
Thinking, streaming, and steering.
That's Wren reasoning before it commits to a reply. On the Grove model, extended thinking collapses into a tidy “Thought for 3s” line you can expand to read the working. The reasoning is there when you want it and out of the way when you don't.
Yes. Auto-scroll is detachable: scroll up mid-stream and Wren stops yanking you to the bottom so you can read at your own pace. A small “Jump to latest” control reattaches you when you're ready.
Group related chats into a Project — see the Projects section below. Within a single chat, everything stays in one thread; start a new chat for a new topic so context stays clean.
You can publish an individual Frame to a shareable link (see Frames). Full-conversation sharing isn't part of this reference build — for now, copy the text you need, or publish the Frame that holds the result.
Live previews, versions, publishing.
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. You read the answer while the thing you asked for renders itself right there. Learn more on the Frames page.
Every edit makes a new version instead of overwriting. Say “make the hero warmer” and you get v2 in place, with a version stepper at the top of the panel. Step back to any earlier version to compare or restore — v1 is never lost. Each version keeps its own Code and Preview.
Open the Frame, choose Publish, and Wren gives you a shareable link to the current version. Send it to anyone — they see the live preview, no account needed. Re-publish to update the link to a newer version. Only the Frame you publish is shared; the rest of your chat stays private.
Try these in order:
the frame is blank, please fix it — it re-reads its own output and ships a corrected version.Grouped chats and standing instructions.
A Project is a folder for related chats plus a set of standing instructions Wren keeps in mind across all of them — a tone, a brief, the facts about your work it shouldn't have to be told twice. Great for a book, a client, a codebase, or a course. See Projects.
Open the Project, click Custom instructions, and write plainly what Wren should always know or do here — for example: “I run a small nonprofit; keep our voice warm and never invent statistics.” Save it, and every chat inside that Project starts with those instructions applied. Edit them any time; changes take effect on your next message.
No. Custom instructions are scoped to the Project you set them in. A chat outside that Project, or in a different Project, doesn't inherit them. Start a general chat when you want a clean slate.
Free includes Projects with fair usage; Pro and up make Projects and Frames unlimited. Wren is a single workspace per account in this build — there are no separate team workspaces to juggle. Compare plans →
Pick the brain and the voice.
Use the model picker at the top of the chat. Choose Seed for speed, Field for the balanced everyday default, or Grove for extended thinking on hard problems (Pro). You can switch mid-conversation — the next message uses your new pick.
Seed is fastest — quick drafts, edits, back-and-forth. Field balances quality and speed for most work. Grove is the most capable, with extended thinking for tough reasoning. Honest note: in this reference build all three are served through the same hosted frontier model with different settings; a production Wren would route each tier to a distinct model. We publish how they score on our evals page.
Open Styles next to the model picker and choose Normal, Concise, Explanatory, or Formal. Style changes the voice and length of replies without changing the model. Concise + Seed is a great fast-drafting setup; Explanatory + Grove is a good study companion.
Your last-used model and Style carry into new chats. For work with a consistent voice, set it once in a Project's custom instructions and every chat there follows suit.
Free tier, upgrades, and limits.
Real Wren, free: Chat, Frames, and Projects, plus the Seed and Field models, with fair daily usage. Grove (extended thinking) and higher limits come with Pro at $17/mo. No card needed to start. See all plans →
Wren tells you clearly — it doesn't fail silently. On Free you'll see a friendly note that daily usage is spent, with the time it resets and the option to upgrade. Your chats, Projects, and Frames all stay put; you just pick up again when the window resets or after upgrading. Paid tiers raise the ceiling: Pro higher than Free, Max 5×–20× Pro.
Pro is $17/mo (billed yearly; $20 month-to-month). Max from $100/mo for 5×–20× Pro usage and priority at peak. Team is $20/seat/mo (billed yearly; $25 monthly) with shared Projects and central billing. Full detail on Pricing.
This is a reference build, so checkout runs through the platform's free placeholder gateway — no real charge and no real card stored. In production these plans would bill normally. We label anything demo-only so you're never guessing.
Your chats are yours.
No. Your chats, Projects, and Frames are yours; we don't train models on them. Wren is a product built on hosted frontier models — not a model lab — so there's no in-house training pipeline your data would feed. The specifics are on our privacy page and data-use notes.
Only you, on your single workspace. The one thing that becomes visible is a Frame you explicitly publish — and only that Frame, at the version you published. Everything else stays private.
Yes. Delete an individual chat or Project from within it, or request full account deletion. What we keep, for how long, and how to remove it is spelled out on the privacy page.
Your messages are processed by the hosted frontier model that powers Wren in order to generate replies — that's how any model chat works. We don't sell your data or hand it to advertisers. Read the specifics, including our processors, in the data-use notes.
When something's off.
Hit Regenerate — a dropped connection is the usual cause and a retry almost always completes. If it keeps cutting off, check the status page for chat-streaming incidents; a proxy hiccup there will show up as a note.
Refresh the tab first — Wren restores your place. If the app itself won't load, glance at status for a Web app incident. As a single web-only workspace, there's no cache to clear beyond a normal browser refresh.
See the Frames section above — the Code tab flags the line, and telling Wren “the frame is blank, please fix it” usually resolves it in one version. A hard refresh clears a stale preview iframe.
Check status for a known incident, then send us a note with what you were doing and, if it's a Frame, the published link. A real person reads these.
We're a small team and we read every note. Leave an email and a sentence or two — we'll get back to you, usually within a day.