AI models
Your favorite AI models, in your chats.
Summon Claude or GPT into any conversation, put more than one to work at once, and let them build on each other — all without leaving the messenger.
New chat
Summon a model with @
How the models work
Call the model you want by name, bring in a second when you need it, and let them build on each other.
- 1
Summon by name
Type @ and pick a model — Claude Sonnet 5, GPT-5.6 Terra, Claude Opus 5. The one you choose answers; mention two in a message and each one replies.
- 2
They build on each other
Reply to what one model wrote and the same model picks it back up. Ask another to weigh in and it responds to the first — a real hand-off, all in one thread.
- 3
Pick your default
Set the model you reach for most as your default, then override it per message with @ whenever another one fits better.
Good to know
One catalog, top models
The Claude, GPT and Gemini families from Anthropic, OpenAI and Google, kept current. No separate apps, no API keys to manage — they are already in your chats.
Better models, higher plans
Every plan can summon a model. The strongest ones carry a plan badge right in the picker, so you always see what a higher plan unlocks before you upgrade.
Summoned, never uninvited
In a group a model replies only when someone @-mentions it or replies to it. It never joins the conversation on its own.
More than one mind on it.
Summon two models in the same message and each one answers. Reply to one and the other builds on it — a genuine back-and-forth, right in the thread.
Questions, answered
The Claude, GPT and Gemini families — the models most people already reach for — from Anthropic, OpenAI and Google, kept up to date. You set a default in settings and can switch to any other model with @ in a single message.
Yes. Summon whichever model fits the moment; mention two in one message and each one replies. A conversation is never locked to a single model.
They build on each other through you. Reply to what one wrote and the same model continues; ask a second to weigh in and it responds to that first answer — a genuine hand-off inside one thread.
Type @ and pick from the catalog — a Sparkles avatar, the model name and its @handle. The one you pick answers. You can also set a default model in settings for when you do not name one.
Every plan can summon a model, and the picker marks the ones above your plan with the plan that unlocks them, so you always see what an upgrade adds. Nothing is hidden — you just choose when you want more.
No. In a group or channel a model answers only when it is summoned — @-mentioned or replied to. It never posts unprompted.