Latenta is a node canvas for AI image, text and video work — connect a prompt to a generator to a preview, press Run, and watch the whole graph fill in.
Generate Image, Edit Image, Generate Video — semantic building blocks. Models and keys live server-side; when a better model ships, your graphs just get better.
Describe the workflow you want and watch it land on the canvas, node by node. Every reply is one undo step — experiment without fear.
Workflows autosave to your library as you edit. Open them from any browser — same graph, same results, cached runs included.
Everything you need to try the canvas — build, wire and preview without limits.
For steady image and video work — enough tokens for daily generation, with top-ups when you need them.
For heavy pipelines — video-scale token budgets for the busiest weeks.