jev supports image inputs now btw this is cool af. had no idea until today it's still in preview, but so many new ideas just became possible i wanna see more people building dope stuff with multimodal decision models https://x.com/weswinder/status/211270339…