MiniMax teased minimax h3 on X with a short post that said, “It’s coming,” and the hashtag #MiniMaxH3. The company also used a teaser video that repeated, “The world is multimodal.” For Users of MiniMax, that reads like a hint that the next model could sit at the center of its AI products.
MiniMax H3 on X
The teaser is the first public signal that MiniMax is preparing what seems to be its next flagship AI model, MiniMax H3. A public tease can prime expectations fast, but it does not tell users whether the model will be faster, cheaper, or more capable in any specific task.
Hailuo AI and multimodal AI
MiniMax tagged Hailuo AI in the post, which links the teaser to its assistant platform rather than treating H3 as a standalone experiment. The company also frames H3 as part of a broader push toward multimodal AI, the kind of system that can handle text, visuals, audio, and video together.
That is the part worth watching closely. MiniMax is teasing H3 as multimodal, but it has not actually said what H3 can do, so the gap between branding and shipping remains wide.
Speech 2.8 and H3
MiniMax has already described Speech 2.8 as having better voice quality, more expressive speaking generation, smoother pronunciation, and more grounded emotional delivery. It says Speech 2.8 can generate speech for AI assistants, audiobooks, podcasts, video narration, customer service, educational content, and digital avatars, which shows the company is already building pieces that could fit into a wider multimodal stack.
For Users of MiniMax, the practical question is whether H3 will stay a teaser or become the model that ties those pieces together. What specific functions will MiniMax H3 actually support when it is released?
