Run powerful AI agents without trusting them with your machine.
§ for your infraAn LLM orchestrator routes tasks. Per-task agents run inside isolated sandboxes — base · browser · ffmpeg · netsec · runtimectl. The Guard authorises every command, every domain, every IP — before it touches a server.
view on github →Gonka is a decentralised inference network — independent GPU hosts serve open-weight models behind an OpenAI-compatible API. No central provider, no subscription; every request is settled on-chain from a GNK wallet.
Gonka is an open-source protocol — contributions welcome across the network: issues, models, GPU compute, integrations.
contribute →$ git clone github.com/x0152/grand
$ cd grand
$ ./quickstart.sh # prints URL + AUTH_TOKEN, then follow the in-app wizard