Octomind Cloud and Hub

One login, zero API keys — cloud agents + 27 models

Productivity

Developer Tools

Artificial Intelligence

GitHub

Visit WebsiteSee on Product HuntTwitter ⧉Github ⧉

Last time I was here, we launched the runtime – one Rust binary, open source, run it yourself. This time I'm here because I got tired of my own agent dying. I'd kick off a refactor, the agent would be forty minutes in, and then I'd close the lid to catch a train. Session gone, context gone, tokens I paid for producing work I'd never see. Agents are long-running processes. Laptops are not long-running machines. So we built the thing I wanted: machines in the cloud that keep running when your laptop doesn't. The part I'm proudest of is sessions – start a turn, close the tab, lose your connection in a tunnel, the turn keeps going and the relay records every step. Open it from your phone later and the full transcript is there. We tested this by killing browsers mid-turn until we couldn't lose a byte. The other thing: I had API keys in five dashboards. OpenRouter, Anthropic, OpenAI, Voyage, a .env file one git add -A away from being public. So the hub side is one login, one key, every model works. 21 of them, including glm-5.2 which solved 24/25 real-PR benchmark tasks – ahead of Claude Code with Opus, and it's in every paid plan. We're bootstrapped, no investors, so we buy hardware as people subscribe. I'd rather hand out machines at the rate we can actually serve them than watch a signup rush turn into "at max capacity" for everyone. The screenshots in the launch post are my actual account spending real cents. That was the bar – not a demo, a place where your agent works while you don't. Ask me anything.