[ryanhecht]

Hi Alex! I'm Ryan, I run a Minecraft theme park recreation server. While our gamemode is very far from vanilla Minecraft (it plays more like a MMORPG), I've been interested in standing up a model like this to help automate agentic testing of new software features in my monolith Bungee/Paper codebase.

In its current state, do you think it can handle non-vanilla Minecraft tasks like interacting with custom UI's, running commands, or doing other unusual tasks in the world?

[agajews]

Probably not the current model, but one of the benefits of doing internet-scale pretraining is that the model has seen a lot of mods already! We think the bigger models will be able to handle some more custom Minecraft worlds.