Voices
fal's post-trained H3 Max renders a 5-second video in under 3 seconds, ~35x the official H3 endpoint
Announced August 27 by fal Research, H3 Max is a post-trained MiniMax H3 with text-to-video and image-to-video endpoints, native synchronized audio, 5 to 15 second durations and an optional final keyframe. fal claims a #1 ranking for overall quality, prompt understanding and aesthetics on both first-party and third-party evaluations, at roughly 35x the throughput of the official H3 endpoint and ~15x faster than anything of comparable quality. Launch pricing is $0.025 per output second at 480p and $0.04 at 768p.
Source
↳ Follow the thread