Open-market AI. Paid by the token. Capped by you.
Independent providers sell their models here. See every price before you ask, pay in USDG for only what you use, and set a limit nothing can cross.
Compare every AI model price before you pay.
Independent providers run the models and set their own prices. Cross AI shows what your request would cost with each one, so you can pick yourself or let a policy like Cheapest or Fastest choose.
“Summarise this 20-page contract.”
- Juno NodesLlama 3.1 8B0.9 s$0.0006
- MercuryGemma 2 9B0.2 s$0.0024
- Vesta LabsQwen 2.5 32B0.5 s$0.0072
- Atlas ComputeMistral Large0.7 s$0.0144Chosen
- OrionLlama 3.1 405B1.8 s$0.0360
Auto reads your message and picks a model that suits it. This request goes to Atlas Compute and costs $0.0144.
Set a limit. It can’t be crossed.
You pay only for the tokens an answer actually uses, at the price shown. If an answer would cost more than your limit, it stops at the limit. Try it.
- 01Set your limitChoose the most one answer may cost. Try the slider.
- 02Ask anythingPick a question and press Send.
- 03Pay per tokenA token is a small piece of a word. Each one has a tiny fixed price, so the cost rises as the answer is written.
- 04Never past itIf the answer would cost more than your limit, it stops at the limit. You never pay more.
Explain how the internet works, in detail.
- Tokens
- 0
- Spent
- $0.0000
- Your limit
- $0.0100
Pick a question, set your limit, press Send.
Already using OpenAI? Change two lines.
Point any OpenAI SDK at Cross AI and use your key. Press Run: every answer says what it cost, and each key can have limits of its own.
from openai import OpenAI
client = OpenAI(
base_url="https://ai.crossmind.lat/v1", # 1. our address
api_key="cm_test_7f3a…", # 2. your key
)
r = client.chat.completions.create(
model="auto", # a model, or a policy
messages=[{"role": "user", "content": "Summarise this ticket"}],
extra_body={"max_cost": "0.05"}, # never more than this
)
print(r.choices[0].message.content)
Give each app its own key and its own limits: per request, per day, per month or in total.
Run a model? Get paid for every token.
List any OpenAI-compatible server, set your own prices, and earn USDG for every token you serve, minus a 2% fee.
$20 a day, after the 2% fee. Paid in USDG to your wallet; withdraw any time.
An estimate: you set the price; how much you serve depends on buyers choosing your model.
Plain answers.
Is Cross AI live?
Yes, on Robinhood Chain with real USDG: the API, the Playground and settlement on chain.
Can I go over my limit?
No. The charge is the tokens used at the posted price, and the contract never pays more than the spending limit you set. API keys can have smaller limits of their own.
Where are my chats kept?
In your browser, saved per wallet. Only the open chat is sent with a request. Cross AI does not store prompts or replies; each listing says whether its provider keeps them.
Who runs the models?
Independent providers. Each listing shows the provider, prices, context size, data policy and measured reliability, so you can choose before you pay.
How are providers paid?
Each reply adds its measured cost to a signed running total, settled on chain in batches. Earnings land in the provider’s balance and can be withdrawn any time.
Do I need the Cross Mind token?
No. Cross AI is priced and settled in USDG only.


