Kandinsky 6.0 Video open-sourced under MIT: 5s clips with 44kHz audio on an RTX 4090 in ~7 min
lmoroney · x · 2026-10-07
Kandinsky Lab open-sourced Kandinsky 6.0 Video under MIT license with code, weights, and Diffusers support. Two sizes—Lite (3B) and Pro (29B)—generate 5-second clips with 44 kHz synced audio including lip-sync from text or an image, and a separate super-resolution model upscales to 1920x1080. Per their timing table, standard-resolution Lite takes 437s on an RTX 4090 and 309s on an RTX 5090 with block offload. In their human evals, Pro trades wins with Veo 3.1 Fast but trails MiniMax H3 and Seedance 2.0 on visuals.
More from Multimodal
- VEDA Sparse Attention cuts MiniMax H3 video gen time in half in ComfyUI with no visible quality loss — robomar_ai_art · 2026-10-07
- Blogger says Opus-generated video matches museum-grade immersive work that once cost $70K teams — oran_ge · 2026-10-07
- Super robot anime made with Kling 4 Flash looks straight out of a studio — Forsaken_Stuff_Ai · 2026-10-07
- Dev builds ComfyUI node pack for projection mapping, 3D region prompts and motion design — Puzzled_Parking2556 · 2026-10-07
- Full Nano Banana 2.1 Prompt for Cinematic Posters That Keep Your Face and Copy the Layout — CodeByPoonam · 2026-10-07
- Nano Banana 2.1 aces classic image-gen traps: full wine glass and a 3:30 clock — _philschmid · 2026-10-07