MiniMax H3 (Hailuo 3.0) Launches with Native 2K Video and Synced Stereo Audio in One Pass
AI video model releases
MiniMax H3 (Hailuo 3.0) Launches with Native 2K Video and Synced Stereo Audio in One Pass
MiniMax H3, also sold as Hailuo 3.0, launched on July 31, 2026 as an omni-modal model generating native 2K video with synchronized stereo dialogue, ambience, and sound effects in a single generation pass, rather than adding audio afterward.
The model is live now via the MiniMax platform API (model ID MiniMax-H3) and the consumer Hailuo AI app, and ranks #2 on the Artificial Analysis video arena leaderboard, with open weights promised "in the coming days."
H3 is built on four new architecture pieces: Contextual Omni Representation, a captioning pipeline that compresses about 100,000 tokens of source material into roughly 4,000 tokens of inference using language to bridge mixed inputs and video output, plus a rebuilt component called H3-VAE.
MiniMax H3 makes native synchronized audio the third instance of a frontier video model shipping this as a default output rather than an add-on, following Kling 3.0 Omni and Gemini Omni Flash within roughly six months.
MiniMax is making aggressive pricing claims against mainstream competitors, positioning H3 as a cost disruptor alongside its technical capabilities.
cho.sh