one key, every model_
OpenAI-compatible.

Make a key on the Account page, add balance with CREDIT, then call the API like OpenAI. Only the base URL and key change.

curl https://agentrose.net/v1/chat/completions \
  -H "Authorization: Bearer $AGENTROSE_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"claude","messages":[{"role":"user","content":"hi"}]}'

import os
from openai import OpenAI

client = OpenAI(base_url="https://agentrose.net/v1", api_key=os.environ["AGENTROSE_KEY"])
r = client.chat.completions.create(model="claude", messages=[{"role": "user", "content": "hi"}])
print(r.choices[0].message.content)

import OpenAI from 'openai'

const client = new OpenAI({ baseURL: 'https://agentrose.net/v1', apiKey: process.env.AGENTROSE_KEY })
const r = await client.chat.completions.create({ model: 'claude', messages: [{ role: 'user', content: 'hi' }] })
console.log(r.choices[0].message.content)

Sign in with your wallet on the Account page and create a key. It is shown once; keep it out of code and git. Send it as a bearer token:

Authorization: Bearer sk-rose-…

Every key spends the balance of the wallet that made it. You can give a key a daily spend cap and revoke it any time.

Use a full id from the list, or a short name that always points at that family’s current model.

curl https://agentrose.net/v1/models

Prices per 1M tokens for every model are on the Models page.

POST /v1/chat/completions takes the OpenAI request body: model, messages, max_tokens, temperature, tools, stream and the rest are passed through.

With stream: true you get server-sent events as usual; the cost arrives in one extra chunk just before [DONE].

Each response carries an agentrose block with what the call cost and what is left:

"agentrose": {
  "used_model": "anthropic/claude-sonnet-5.5",
  "cost_usd": 0.000412,
  "saved_usd": 0,
  "balance_left": 9.87
}

Calls are billed at list price from your balance. The discount comes when you buy that balance: $1 of it costs $0.60 in CREDIT.

Building with Claude Code, Codex or another coding agent? Copy the setup prompt and paste it in. It tells the agent where your project calls an LLM, to switch the base URL and key, and to send one test call.

1 CREDIT = $1 of AI. It is a token on Robinhood Chain, paid in ETH.

Contract addresses and how the money splits are on the Protocol page.