Video Rebirth just released HyperFlow H3 Flash, an open ~14B distillation of MiniMax-H3 that generates synchronized 768p video and audio in 4 to 8 steps. The Q4 GGUF weight drops down to 9.1 GB to run on 24GB RTX GPUs, while 4x Blackwell B200s achieve playback-speed real-time generation. Why it matters: Cutting-edge, synchronized audiovisual models are finally breaking out of hyperscaler clusters and onto local creator hardware.
Pruna AI just released P-H3-Image, an open-weights image model built by reworking the MiniMax-H3 video model. By distilling components and slashing diffusion steps from 50 to 8, it runs 5x faster and shrinks down to 20 GB while rendering up to 16 megapixels. ComfyUI already supports it natively.
Pika just open-sourced Pika H3 Flex, a post-trained model built on MiniMax-H3 that generates videos with synchronized stereo audio in only six sampling steps. It handles 5-to-15-second clips across 11 aspect ratios, accepts up to 12 multimodal references, and runs on a single H100 GPU. Why it matters: professional creators get rapid, high-control iteration without waiting minutes for a render.
Nvidia just launched Open H3 Regenerate 2K, an open-source tool that redraws low-resolution video drafts at 2K and up to 4K in just three diffusion steps. Built primarily for MiniMax H3 but compatible with models like LTX-2, it gives creators full local control over high-res video refinement, though it may slightly alter fine facial details.
FastVideo just released Fast H3 Omni, an open-source acceleration method for MiniMax H3 that cuts sampling steps from 50 to just 8. Built specifically for reference-to-video workflows to keep characters consistent, it slashes latency by up to 12x on top-tier GPUs. Ready-to-use checkpoints for ComfyUI, FP8, and Apple Silicon are already available on Hugging Face.
MiniMax just launched M3.1 Flash, the official release of its anonymous model previously tested as 'Space Bunny.' Tuned for coding and computer-use agents, it runs nearly three times faster while slashing usage costs by 70% to $0.15 per million input tokens. Weights will follow, but API and desktop access are already live.
Researchers just dropped Any Time, Any Space (ATAS), an open-source framework solving scene drift in long-form AI video. By reconstructing scenes in 3D and storing dynamic spatial memories, cameras can leave a room and return without anything changing. It plugs into models like Wan and MiniMax, with code ready to run locally.
Nvidia just open-sourced Stream Render H3, a real-time diffusion system built on MiniMax H3 that replaces traditional game rendering. The engine only tracks physics and outputs flat semantic maps, while the AI generates textures, lighting, and customizable visual styles. Catch is, running it live takes 8 enterprise GP200 GPUs and over 200GB of storage.
ByteDance introduced DMAD, an adversarial distillation method for rapid visual generation. By training a student model to match real data distributions using a shared-backbone discriminator, it compresses 20-to-30-step video diffusion models like MiniMax-H3 down to four denoising steps while preserving fidelity.
Developers just open-sourced Mosaic Chunk, a memory system that helps AI video generators track objects and scenes long after they leave the frame. By selectively storing key features instead of every single frame, it maintains visual consistency across cuts. The release includes local-ready models based on MiniMax H3 and Ling World Infinity.
NVIDIA just introduced SoL-Refiner, an AI model that transforms low-resolution video drafts into crisp 4K outputs using a single denoising step. It works across different base generators, slashing MiniMax H3 generation time by 27x down to under six seconds. While it supercharges high-res rendering, it does alter or reimagine some fine details from the original clip.
Researchers just open-sourced PDMD, a video distillation technique that shrinks MiniMax-H3 generation from 30 steps down to just four. By curbing misleading critic feedback during training, it keeps motion smoother and visual details sharp instead of muddying the frame. Fast open-source video generation is getting shockingly usable.
Viggle AI just launched Meridian, a geometry-guided video model built on MiniMax-H3 that re-renders existing footage from new camera paths. You can feed in a standard video or single photo, estimate its 3D depth, and generate smooth orbits, pans, or frozen bullet-time moments. The model is available as two lightweight LoRAs alongside dedicated ComfyUI nodes.
Viggle just open-sourced Viggle-Animate, an AI pipeline built on MiniMax-H3 that ditches traditional pose skeletons, masks, and face trackers. You simply edit one frame of a video to introduce the new character, and the model propagates it across the entire shot in three forward passes, rendering 124 frames in 26 seconds on a single B200 GPU.
Hao AI Lab and MiniMax just released FastH3 v1, an open-weight video generation model that creates a 15-second 768p video in only 13 seconds. Clocking up to a 14x speedup on NVIDIA Blackwell GPUs with a single flexible checkpoint, it brings production-grade speed to the open-source community. Full weights and inference code are out now.
Researchers just open-sourced Code World Model, decoupling world logic from visual rendering. Instead of merely predicting the next frame's pixels, a coding agent acts as the 'brain' to update persistent scene states in code, which a fine-tuned MiniMax-H3 video model then renders. Why it matters: It stops interactive AI worlds from forgetting objects, keeping rules and consequences coherent over long horizons.
BreezeBlue just released open weights and inference code for Breeze-TTS-2, a bilingual text-to-speech model built for real-time interaction. It delivers sub-40ms time-to-first-audio on an H100 and lets developers design distinct voices from scratch using natural language prompts or inject vocal cues like sighs and laughs directly into transcripts. The code is Apache 2.0, while model weights remain research-only.
MiniMax just launched Music 3.0, an open-weights AI model that generates complete songs up to five minutes long in a single pass. Powered by a hybrid architecture based on Qwen3.5-8B alongside flow matching, it preserves long-range song structure while capturing realistic instrument dynamics. Why it matters: It brings fully produced, extended AI tracks into the open-source fold without sacrificing acoustic fidelity.
MiniMax just unveiled H3, an omni-modal generation model that natively handles text, images, video, and stereo audio in a single context. It generates up to 15-second clips at 2K resolution at a fraction of competitors' costs. MiniMax also plans to open-source the model weights in the coming days, taking a rare open-weights bet against closed video giants.
MiniMax just released M3, an open-weight model packing a 1M token window, native multimodality, and desktop computer use into a single package. Its custom sparse attention architecture cuts decoding compute dramatically, yielding a 15x speedup over its predecessor and top marks on SWE-Bench Pro. It brings closed-source agentic capabilities straight to open-weight developers.
