ChainThink reports that on September 3, according to an announcement by fal, the AI inference platform fal launched its Turbo version just one week after the release of MiniMax H3 Max, reducing generation latency by approximately 2.5 times and halving the price.
The Turbo version is post-trained on MiniMax H3 and optimized for the fal inference system. In official tests with 5-second, 768p videos, generation latency decreased from 3.49 seconds on H3 Max to 1.40 seconds.
In terms of quality, Fal's self-tested Turbo version achieves 97% of the original H3 Max, with prompt understanding, aesthetics, and overall quality largely preserved, remaining superior to the original MiniMax H3 and other tested models.
In terms of pricing, Turbo is normally $0.04 per second, which is only half the regular price of H3 Max ($0.08 per second). During the launch promotion, it is $0.01 per second, costing approximately $0.60 to generate a 1-minute video.