alt Hacker NewsFrom README:
> On the 128 GB M5 Max, clean end-to-end image+audio and embedded-video+audio renders completed in 74.58 and 76.99 seconds respectively, each with about a 40.1 GB peak physical footprint and zero swaps.
Looks like it uses 40GB? So your 96GB mac setup should work fine i guess (Model itself is 33B)
This repo looks neat, but I hope they add some more clear benchmarks because that time (74.58s) is pretty meaningless given that the it/s (and total time) is highly dependent on mode (T2V vs I2V vs REF2V), resolution (0.4, 0.6mp, etc), duration (5-15 seconds), etc.