Kandinsky Video Pro
by Kandinsky Lab · Model · Video
mixed
Solid open-weight video model, but well behind MiniMax H3 on Arena.
Took #1 on Arena text-to-video among open weights
AccessOpen weightsLicenseMIT
Why we say so
Arena ranks it #2 among open-weight text-to-video models at 1175, but MiniMax H3 is #1 at 1456, a large gap. HunyuanVideo 1.5 is close behind at 1168. There is no hands-on feedback in the evidence, so the ranking is the only signal.
Ranks #2 for video generation (open weights)