The plan decides how the GPU is allocated. Heavy+ adds an X agent on an X account you own.
GPU allocation
Starter vs Heavy
| Starter | Heavy | |
|---|---|---|
| GPU allocation | One GPU across up to 4 projects | 100% of the GPU, no sharing |
| Speed | Under 15s, up to 30s at peak | 5-8s per image |
| Throttling | Round-robin queue at peak | None |
X agents
An X agent runs on an X account you own. It watches that account for mentions and replies to a mention with a generated image.
A mention gets a reply only when the sender is on the bot's whitelist and the mention contains the account's @handle together with a character trigger word. The agent only replies; it never posts on its own.
You supply the account and the API keys
The X account and the X Developer API keys are yours. You enter them at deployment and the X API bill goes to X directly.
Heavy+ — $499/mo
| Setting | Value |
|---|---|
| Compute | Heavy (dedicated) — 5-8 seconds per image |
| Telegram bot | A matching Telegram bot on the same character, included |
| Hosting | Server provisioning and upkeep of the live agent |
| Generations | No cap, no per-image fee |
The whitelist is the Whitelist tab in the dashboard: bulk-paste only, up to 200 handles.