MiniMax H3 omni-modal video generation in ComfyUI. Text/image/video/audio in, video with native 32kHz stereo audio out (768p default, 1080p+ supported). Disk-optimized: pruned INT8 + NVFP4 weights (~63GB instead of ~290GB). NVIDIA only.
Fish Audio Drama 3 (expressive TTS, voice cloning, voice design, directing notes) from the fishaudio/Drama-3 Space. Models: the full model for 24+ GB NVIDIA GPUs, the Small GGUF (drama3-small, Japanese and English) for 16 GB, and MLX 4-bit / 3-bit conversions of the full model for Apple Silicon (64 / 48 GB, untested on a Mac).
Speridlabs Iris-3B pixel-space diffusion: text-to-image, monocular depth, and image restoration / 4x upscaling in one local Gradio UI. NVIDIA GPU recommended (16 GB+); Apple Silicon and CPU work but are slow.
[LINUX + NVIDIA ONLY] Real-time interactive world model. Drive an infinite, action-conditioned world rollout at 720p/16fps on a single desktop GPU (~19GB VRAM). https://github.com/amap-cvlab/ABot-World
