These docs cover RhinoArtisan 7.0, which is still in beta. The contents are provisional: features and pages can still be added, changed or removed before the final release.
AI Models & Services in RhinoArtisan
Different Models for Different Jobs
A model learns from a particular kind of data and can only work in that domain. A language model has never seen a rendered ring; an image model has no concept of carat weight. So the tools in RhinoArtisan reach for different models depending on what you asked for.
What each job needs
Language and reasoning: the Assistant and Flow Studio AI. Answering a question about a setting, or turning a description of a repetitive task into working code. These call large language models, and code generation in particular rewards a strong one.
Images: Generative AI. Turning a prompt or one of your own designs into a photographic render, a sketch, a variation or a campaign image. A completely different family of model, trained on pictures.
Video and 3D: also Generative AI, and different models again. Turntables, motion, and reconstructing geometry from an image are each their own specialty.
Why two providers, not one
No single company makes the best model in every category. The strongest language model and the strongest image model come from different labs, and the ranking changes several times a year.
Rather than tie you to one vendor’s whole catalog, RhinoArtisan connects to two gateways that each aggregate many providers: OpenRouter for language, and fal.ai for images, video and 3D.
The practical effect is that when a better model appears, it shows up in your list. You are not waiting for us to ship an update, and you are not locked to whichever lab we happened to sign with.