GitSeonjeap
English

Sign in

Reading needs no sign-in. Sign in to write posts or comments, or to recommend and report, in the community.

By signing in you agree to the notices below. Community rules · Privacy notice

Keep browsing without signing in

← News

Video generation

MiniMax H3-Image Open Weights Released, Image Generation and Editing Model

Posted As of

MiniMax has released the H3-Image model card on Hugging Face, introducing an open-source model for image generation and editing. The official Model Zoo table lists file sizes for the BF16 DiT and the optional prompt expander.

What happened

  • Source: MiniMaxAI Hugging Face model card (MiniMax-H3-Image)
  • Card description: an open-source model supporting image generation and editing, part of the MiniMax H3 family
  • Model Zoo table (official): BF16 DiT 68.65 GB, optional prompt expander MiniMax-H3-Image-PE 17.9 GB (NVFP4)
  • The four execution paths (SGLang, vLLM-Omni, diffusers, ComfyUI) are based on third-party mirror reports; not confirmed in the official snippet
  • Weight upload time 2026-11-10 14:50 UTC is based on third-party mirrors; the card snippet has no date
  • Parameter count is not in the official snippet. The 33.1B figure comes from a community GGUF card
  • Some press reported that the license excludes the US, EU, and UK, but this was not confirmed in the official card

Our take

  • The BF16 DiT file is listed at 68.65 GB, so local execution appears to require substantial storage and memory headroom.
  • Unconfirmed points: parameter count, whether the license has regional restrictions, and actual behavior of each execution path. Regional availability cannot be determined until the license file is checked directly.

Sources and related posts

More in Video generation

  1. NVIDIA Cosmos 3 Four Open Checkpoints for Driving and Robot World Simulation (huggingface.co)

    NVIDIA has added four open-weight checkpoints for interactive world simulation to the Cosmos 3 family. They target multi-sensor driving scenes and bimanual robot simulation, and are not general-purpose text-to-video models.

    Video generation| |1 sources |Our take included
  2. Google Gemini Omni 1.2 Flash Generally Available, Supports 15-Second Videos (blog.google)

    Google announced that Gemini Omni 1.2 Flash is now generally available on the Gemini API, Google Cloud and more. It supports 15-second video generation and multi-turn editing, and the 720p price is $0.07 per second (official).

    Video generation| |2 sources |Our take included
  3. LTX-2 Code v1.6.0, Code Moves to Apache 2.0… Weights Keep a Separate License (github.com)

    Lightricks updated the LTX-2 repository to v1.6.0 (November 4) and v1.6.1 (November 5). The code is now Apache 2.0, but model weights are not covered by that license, and several legacy APIs were removed.

    Video generation| |1 sources |Our take included

All news