Learn about this first
Pika Labs post-trained on MiniMax's H3 model and released and open-sourced the Pika H3 Flex video generation model. According to X user Khazix (@Khazix0918), the model compresses the sampling steps for video generation from 50 to 6, greatly reducing generation time while even improving quality.
The performance claims above come from a description relayed from Pika; the report did not provide independent testing or comparative data.
AI-generated from reports · updated 18 days agoLatest developments
Pika Post-Trains on MiniMax H3 and Open-Sources the Pika H3 Flex Video Generation Model
Event progress1
Pika Labs open-sources the Pika H3 Flex video generation model
Khazix: Pika post-trained on MiniMax H3 and open-sourced the Pika H3 Flex video generation model