docs
How it works
A compute launch is one transaction. You choose a model class, name the coin, and describe how it should behave. The token deploys and the persona is wired to its mint address.
From then on the coin has a public page on the pulse board and a private chat room gated by holdings.
model classes
- GPT classGeneralist. Fast, agreeable, always online.
- Claude classLong memory, careful tone, writes essays.
- Gemini classMultimodal brain, reads charts for breakfast.
- Llama classOpen weights energy. Loud and unfiltered.
- Mistral classSmall, sharp, European. Ships fast.
- DeepSeek classMath head. Reasons out loud in public.
- Grok classTerminally online. Replies in memes.
- Qwen classPolyglot. Answers in any language you throw.
questions
- What does compute actually do?
- It bundles a token launch with a persona. You pick a model class and write the personality, and the coin ships with a chat attached to it.
- Who can talk to the persona?
- Any wallet holding 10,000 or more of that token. Below the threshold the chat stays locked.
- What does it cost?
- Network fees and the launchpad's own fee. compute does not add a cut and does not take a dev allocation.
- Can I change the model later?
- The model class is fixed at launch so holders know what they bought. The personality prompt can be refined by the creator.