Where it runs
Lodestone uses only fetch, AbortSignal and URL, so it runs wherever your function does. This page says, for each platform, how much of that has been seen and not just assumed.
Netlify
KeyStone calls AI models through Lodestone from Netlify Functions, with the 30 second function limit and a 24 second time budget.
Shared state: netlifyBlobsStore
Cloudflare Workers
The checks run inside workerd, Cloudflare's own runtime, with no Node compatibility flag, on every change. Not deployed to a live Worker yet.
Shared state: cloudflareKvStore for the catalogue and cached answers; Upstash for counters
Base44
Base44 says its backend functions run on Deno, and the checks run on Deno. Not deployed on Base44 yet. Import it with the npm: prefix.
Shared state: upstashStore over HTTP, or memory
Vercel
Vercel's Node.js functions run the same code the checks run on Node 20, 22 and 24. Not deployed on Vercel yet, and its Edge runtime has not been checked separately.
Shared state: upstashStore over HTTP, or memory
Render
A Render web service is a Node server, covered by the Node checks. Not deployed on Render yet. A server has no function time limit, but the time budget still protects the person who is waiting.
Shared state: upstashStore, or memory on a single instance
Runtimes checked on every change
Logos are from the Simple Icons set (CC0) and belong to their owners; they are here to say where Lodestone runs, not to imply endorsement. Base44 has no mark in that set, so it appears as a name.
"Runtime tested" is a claim about the runtime, not the platform: a platform can add limits of its own, such as a smaller time limit or no writable disk. It moves to "In production" only when something real is deployed there.
Deployed it somewhere not listed, or on one of these? Tell us in an issue. More detail: Runtimes, Stores, the Python port.