What we've been building

Click any input node label to rename it. Image, video, and text input nodes now support custom labels for better board organization.

Connected video inputs now display frame thumbnails directly inside model nodes — so you can see what's wired up at a glance.

Image and video inputs now automatically detect their native aspect ratio on upload and on load. Nodes resize to match your media.

Increased the polling timeout to 20 minutes for slow video models like Kling Motion Control Pro, preventing premature timeout errors.

Three new utility nodes for post-processing. Inpaint to edit regions, outpaint to extend, and crop to reframe — all connected in your workflow.

Extract individual frames from video nodes and feed them into image workflows. Bridge video and image pipelines seamlessly.

Two new state-of-the-art video generation models are now available on the canvas. Generate cinematic clips from text or image prompts.

Your board now saves automatically as you work. No more lost progress — every change is persisted in real time.

Upload reference images directly onto the canvas and annotate outputs before sharing with your team.

Usage-based token billing is now live. Each model has transparent per-generation costs with a minimum 17% margin guarantee.

Full undo and redo support with snapshot-based history. Up to 50 undo steps with keyboard shortcuts.

Select multiple nodes with Shift+click or box selection. Move, delete, or organize groups of nodes at once.

Browse all your generated outputs in a gallery view. New flow variant picker for quickly switching between model workflows.

Preview your generated images in Instagram, TikTok, and LinkedIn post templates before exporting.