# Best open-weights AI model for video generation — October 2026

As of Oct 15, 2026, Wan V2.2 A14b is state of the art in Video generation, open weights (since Aug 7, 2025; sources LMArena Text-to-Video; methodology v1.0).

Leading, crown pending confirmation: Kandinsky 5.0 T2v Pro.

Category board: https://isitsota.com/c/video-gen?slice=open-weights

## Top 10

- #1 Kandinsky 5.0 T2v Pro
- #1 Wan V2.2 A14b
- #3 Kandinsky 5.0 T2v Lite
- #4 Mochi V1

## Crown changes

- Aug 7, 2025 — Video generation: Wan V2.2 A14b takes the crown

## Weights

- LMArena Text-to-Video: primary, weight 1