Last week I shared a simple server for using Metappuccino, and so far my work is happy to let me keep using this old GPU machine to host various bioinformatics tools, so I've now stood up a very basic, free-to-use Boltz2 server. You can find it at boltz2.biobench.ai

Why do we need another Boltz2 server? I was seeing some people not able to use a lot of the existing free ones because they limit the size of your proteins, require trial account signups, or limit how many runs you can do. I've taken a different approach - you can submit whatever you want whenever you want, the only catch is that I only have one GPU for this so if someone else is using it you have to wait, and any job you have in the queue needs an occasional live button click to encourage people to take turns.

The server just runs MMSeqs (locally, so I'm not hammering the public MMSeqs server!) and then boltz2 predict <yaml_file> for any yaml you upload. If you need more than that please message me though - my goals here are to practice web server skills, not make the ultimate boltz2 experience, but I'm still keen on making actually useful tools so happy to hear suggestions or other features you might want.

Same story as the Metappuccino server (which is at metappuccino.biobench.ai, if you need that); I'm asking for an email address only as a janky captcha, and I guess to have the possibility of recovering some job logs/results if you do something really important here and then don't download it.