MiniMax H3 Local
MiniMax's H3 video model running offline, with synchronised audio. The heaviest local entry by a wide margin: its text encoder alone is a 32B vision-language model.
MiniMax's H3 Local is an open-weight video-generation model. It produces short video clips from a text prompt. It runs entirely on your own hardware through GGML: no API key and no per-use fees.
- Released
- Aug 2026
- Pricing
- Free, runs on your hardware
Samples
Generated with H3 Local via GGML. The same two prompts run against every video model in the catalog (one landscape, one portrait), so the only thing that changes between two models’ clips is the model.
Camera motion, scene coherence over time
Fluid motion, fine detail, temporal stability
MiniMax
MiniMax is a Chinese AI lab whose speech and music models are widely used for multilingual voiceover and generative music.
www.minimax.io ↗MiniMax's low-cost coding and agent model with a 1M-token context and tunable thinking depth, on Runware and OpenRouter.
MiniMax music model for full-arrangement AI compositions with rich instrumentation.
MiniMax's studio-grade speech model with 332 voices, emotion control, voice cloning, and a 50,000-character input limit.
MiniMax's Hailuo 3 video model with native synchronized audio, multi-reference consistency, and prompt-driven editing of a finished clip.
MiniMax's throughput-tuned H3, built for faster clips with native synchronized audio and first-to-last frame control. It trades H3's multi-reference conditioning and 2K tier for speed.
A faster, lower-cost tier of MiniMax's H3 Max for text-to-video and first-to-last frame clips of 5 to 15 seconds with native audio, at 480P or 768P.