Using TALOS
Everything you need to start chatting, pick the right tier and keep track of what you're spending.
Getting started
- Go to usetalos.xyz and sign in.
- Pick a tier: Light or Heavy.
- Type a message. A node picks it up and the reply streams back.
Picking a tier
| Tier | Model | Cost | Runs on | Notes |
|---|---|---|---|---|
| Light | Nimbus 8B | 8 credits | Browser node | fastest replies, good default |
| Heavy | Atlas 30B or Atlas Vision 27B | 14 credits (18 with deep-think) | Rig node | pick a model from the chat picker; vision + live web lookups |
Heavy gives you a model picker: Atlas 30B for straightforward reasoning, or Atlas Vision 27B when you want image input or the model to check the live web. Either one can also run in a slower, deeper "think" mode for harder problems; that mode costs a bit more per message.
Credits
- 1 credit = $0.01, topped up with USDC from the app.
- Credits never expire.
- A failed job is refunded automatically: you're only charged for a completed reply.
- If you're staking $TALOS, your daily free-credit allowance is drawn first, before anything you've paid for. See Staking.
Web lookups and vision
On the Heavy tier, the model decides for itself when a question needs current information and calls a lookup tool: you don't have to toggle anything. Drop an image into the chat and Atlas Vision will read it directly. Both features are exclusive to Heavy; Light is text-only and works from what the model already knows.
Generating images
Head to usetalos.xyz/create for image generation, a separate flow from chat, with its own pricing and controls. See Image generation for the details.
Privacy while you use it
Prompts pass through the relay and are discarded: they aren't logged against your account. The node that runs your job sees only the text it's asked to process, never who sent it.