Materials, capabilities and models have different jobs. Keeping them separate lets you build workflows around your task, reuse their results and change implementations as you go.

The official tongflow-modal-minimax-h3 plugin brings the best open-source omni-modal video model to the canvas, with kijai's Sol-Attn sparse attention built in. A 5-second clip now renders in under 3 minutes — no extra setup required.

TongFlow is now open source. The first public release, v0.1.0, ships the desktop app, the plugin SDK, and the official plugin catalog — everything you need to build multi-modal GenAI workflows on one canvas.

Treat every AI model as a modality transform, and orchestrate your ideas with just three actions — Add, Transform, Combine — on one infinite canvas.