An OpenAI-compatible API that records what it served

Same client, one base URL

The surface mirrors the conventions your stack already speaks — /v1/chat/completions and /v1/models — so existing OpenAI clients point at Basis with a base-URL change and a sk-basis- key. Streaming uses standard server-sent events. What is different is underneath: accounting is deterministic, receipts are hash-stable, and a worker-routed job rewards the contributor GPU that served it.

curl — a streaming chat completion

curl https://basis.watch/api/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-basis-..." \
  -d '{
    "model": "basis-default",
    "messages": [{ "role": "user", "content": "Summarize this transaction." }],
    "stream": true
  }'

OpenAI SDK — set baseURL, keep your code

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://basis.watch/api/v1",
  apiKey: process.env.BASIS_API_KEY, // sk-basis-...
});

const stream = await client.chat.completions.create({
  model: "basis-default",
  messages: [{ role: "user", content: "Summarize this transaction." }],
  stream: true,
});

for await (const chunk of stream) {
  process.stdout.write(chunk.choices[0]?.delta?.content ?? "");
}

List the catalog

curl -s https://basis.watch/api/v1/models

Pay, reserve, receipt

The public API is a free, metered demo today — it does not require $BASIS credits. The accounting model it runs on is the real one: usage is metered deterministically per token in integer base units, floor division, no floating point in the money path. The payment router can quote ETH, WETH, or USDC into $BASIS at request time — quotes only, with your wallet signing every swap; paid $BASIS credits come online behind their own gate.

Why receipts matter

Model catalog

The multiplier scales pricing and rewards relative to 1x. Each tier in the catalog is served through the live OpenAI-compatible runtime.