MiniMax H3 Emerges as a Strong Open-Weight Video Model

MiniMax H3 is live on Runway: The new open-weight video model is now accessible inside Runway's platform alongside all other frontier models. H3 specs: Native 2K (2560x1440) resolution, up to 15-second clips at 24 FPS, synchronized stereo audio, and character consistency across scenes. Multimodal inputs: A single H3 request accepts up to 9 reference images, 3 video clips, and 3 audio tracks for combined context generation. Cost efficiency: A 15-second 2K clip costs roughly $1, and H3 ranks #1 on Artificial Analysis for video editing with audio. Runway's platform play: Runway now bundles Seedance 2.0, Kling 3.0, Sora 2 Pro, and more alongside its own Gen-4.5, positioning itself as a multi-model creative hub. [...] Open weights: H3 weights are publicly available for self-hosting, private deployment, and fine-tuning via the MiniMax platform. [...] AlphaSignalAlphaSignal # Runway Absorbs MiniMax H3, Betting on a Multi-Model Creative Platform MiniMax H3, a 2K open-weight video model w

### Green-Screen Fairytale Composite "Remove the green screen background of Video 1 and turn it into a fairy tale-like background similar to Video 2. The background elements need to completely match the actions of the characters in Video 1. Modify the lighting of the characters in Video 1 so that it completely matches the background." ## Generate, reference, and edit Create videos from text, control the first and last frame, combine images, video, and audio as references, or edit existing footage with natural language. MiniMax H3 Text to Video MiniMax H3 is a frontier video model. This endpoint generates video from a text prompt alone, rendering at 2K in durations from 5 to 15 seconds across seven aspect ratios. MiniMax H3 Image to Video [...] MiniMax H3 generates 5 to 15 seconds at 24 FPS. Output is 2K, which puts 1440 pixels on the short edge for ratios between 16:9 and 9:16 and reaches roughly 3.7 megapixels on wider formats, for example 2976x1248 at 21:9. Text-to-video and re

Aug 3, 2026 MiniMax releases H3 video model weights, putting an open model at the top of a video ranking for the first time. Artificial Analysis ranks H3 first in Video Editing, second in Text-to-Video, and third in Image-to-Video. The 33-billion-parameter model processes text, images, video, and audio together, generating four- to 15-second clips with stereo sound. According to the model card, a single prompt can include up to nine reference images, three video clips, and three audio clips. Video by MiniMax H3 [...] Video by MiniMax H3 Two pieces remain closed, though. The 2K resolution module and H3-Context-IR, which translates prompts and reference material into a structured intermediate format, aren't included. Running H3 locally in ComfyUI tops out at 768p, and users will need to handle context prep themselves using MiniMax's published prompting guides. The open weights do allow fine-tuning on custom footage, characters, or a specific visual style. One catch on the license side

4,143 followers BREAKING: MiniMax M3 by MiniMax is #10 overall on Design Arena with an Elo of 1320. Debuting with frontier-level performance in coding and agentic work, M3 is the first natively multimodal open-weights model supporting both image input, video input, and desktop computer operation. MiniMax M3 is a 35 Elo point improvement from MiniMax M2.7, putting it in the same performance band as Claude Sonnet 4.6 by Anthropic, GLM 5 Turbo by Z.ai, and MiMo-V2.5 Pro by Xiaomi Technology. Compared to MiniMax M2.7, MiniMax M3 improves the most in the 3D Design, Blog Website, and Portfolio Website categories. Additionally, MiniMax M3 establishes a new Pareto frontier overall on Design Arena for Preference vs Price. Huge congrats to the MiniMax team on the launch!

On top of that, it lands in the top three for text-to-video and the top three for image-to-video, going head-to-head with Google and ByteDance. Now, I'll be straight with you because I always am. It's number one in editing, not number one at everything. In text-to-video, it sits just behind Google's model, and in image-to-video, it's behind ByteDance and Google. So, beat the biggest models is true where it counts most, editing, and it's right in the mix everywhere else. For a model that also plans to open up its weights, that's still a very big deal. And that's the other piece. Minimax says it's releasing H3's weights under its own community license. In plain English, the model itself gets opened up so people can download it, study it, and build on it, not just reach it through a locked [...] this is for you. If you edit and recut a lot and you're tired of regenerating from scratch every time, this is for you. Creators making ads, short films, product videos, social clips, game concept

Kamryn Ohly Intelligence @KamrynOhly 23h A defining moment for open weights in video generation! user avatar Alice The Ai Expert @AliceInfoAi 9h MiniMax H3 just set the bar for open video. Huge leap. [...] Log inSign up ## Post user avatar Design Arena Intelligence @DesignArena MiniMax H3 by @MiniMax\_AI is 2nd overall on Video Arena with an Elo of 1325. This is a 209 Elo increase from @MiniMax\_AI’s previous video model, MiniMax Hailuo-2.3 (Pro), putting them behind Gemini Omni Flash by @GoogleDeepMind and ahead of Seedance 2.0 Mini by @BytePlusGlobal. With this performance, they establish themselves as the 2nd video lab overall. Among open weights, the model is 1st overall, ahead of LTX 2.3 by @Lightricks and Kandinsky 5.0 Pro by AI-Forever. By a substantial gap, MiniMax has set a new SOTA on open weight video generation. Congratulations to the @MiniMax\_AI team on the achievement! 10:06 PM · Aug 4, 20267.9KViews user avatar Kamryn O