Open source plea: Closed video models hinder adoption

VoidAsuka · x · 2026-09-02

The author argues that use cases won't普及 if SOTA video models remain closed-source (citing Seedance 2.0 and H3). With the release of Fable 5.1, there's an urgent push to use tokens now, fearing restricted access to SOTA LLMs and harder distillation later. The post cites a demo of a generative video classroom built with fal H3 Max to advocate for open source.

Original post →

More from Fun

Fun channel →