Pay for the room, not the models.
The plan is what the group does together — the discussions, the rounds, the receipts, the sync. The models run on whichever keys or subscriptions you choose. On your own keys or CLIs, that is all you pay. On ours, you pay what each call actually cost plus our margin, and the receipt shows it per seat.
What a month actually costs
$19
Pro, for the product. Every discussion, every check, every review.
+ $0 to ~$25
For the AI. $0 if you already pay for Claude Code, Codex or Gemini CLI — those seats run on your existing subscription. Otherwise a few cents per turn on your own keys or ours.
≈ 20–100
Discussions a month for a heavy daily user. The app shows the running cost beside the send button, and lets you cap it.
Try the group chat
Includes $2 of AI to start. After that, bring your own keys or top up.
For serious daily use
For the product. The AI itself is separate — your own keys, your own CLI, or ours at cost + 30%.
Per seat, for teams
For the product. The AI itself is separate — your own keys, your own CLI, or ours at cost + 30%.
Prices in USD, billed monthly, cancel anytime.
The models cost money to run. Pick who pays for them — this is separate from your plan, and switching takes one click.
Your keys, stored encrypted, used to call the providers on your behalf.
Claude Code, Codex, Gemini CLI or Grok Build, running on your own machine under your own sign-in.
Nothing to set up. We call the providers for you and bill it to your credit balance.
Where your text goes depends on the app, not only the option. The desktop apps call the providers directly with your own key or your own CLI, so nothing passes through us — though the provider still receives your prompt, as it must to answer. On the web and iPhone the discussion runs on our servers whichever option you pick — with your own key we hold it encrypted and make the call for you. “Our keys” means we pay for that call as well.