中文 Get for macOS

Requirements · Install

Requirements and install

v0.1.1 is a macOS Apple Silicon (arm64) desktop app. This page lists only the published requirements: chip, disk, install steps, and where files go. Minimum macOS version and memory are on GitHub Releases.

Machine Models Disk Install steps Where files go FAQ

AI summary · quotable

An Apple Silicon Mac, plus generous free disk

BigBotArt needs an Apple Silicon (arm64) Mac. The v0.1.1 release notes list M1, M2, M3, and M4. Models and compute run on that machine.

Video uses h3.c MiniMax-H3 (~144GB); images use mlxgen (mlx-gen-jzg) models downloaded on demand. Allow tens to well over a hundred GB of free disk before the first run. macOS version and memory figures stay on the GitHub Releases page.

There is no account and no subscription. Projects, the canvas, takes, and assets are written only to the local directory ~/.bigbotart.

Machine

Machine

There is one desktop build. The site does not list a build for any other system.

Apple Silicon (arm64)

You need an Apple Silicon Mac. The release notes say BigBotArt-0.1.1-arm64.dmg is for M1, M2, M3, and M4.

Compute stays on the Mac

h3.c (MiniMax-H3), mlxgen (mlx-gen-jzg) image models, and any OpenAI-compatible local LLM endpoint run on that Mac. Final export runs through local FFmpeg. See “Local models and engines” below for the full list.

OS version and memory

The exact minimum macOS version and memory requirements are on GitHub Releases: Releases. This page does not add numbers.

Models · Engines

Local models and engines

This list matches the bigbot.art source of truth (h3.c, mlxgen package mlx-gen-jzg, OpenAI-compatible Agent). Sizes are approximate; models download on demand — you do not need every weight at once.

Video · h3.c

MiniMax-H3 (BF16, ~144GB): shot-level video with audio; text-to-video / first-last frame / Ref2VA. Optional style packs: merge LoRA offline into a copy of the base model directory, then point inference at that copy.

Agent · Export

Chat uses any OpenAI-compatible endpoint (for example LM Studio); it is not pinned to one fixed weight. Final export runs through local FFmpeg. Optional post-process: Real-ESRGAN / GFPGAN / CodeFormer (install separately on the Mac).

Image · mlxgen (mlx-gen-jzg)

Default text-to-image / latent img2img uses Z-Image-Turbo; reference images use Qwen-Image-2.1; inpaint and outpaint use Qwen-Image-Edit or FLUX.2 Klein; upscale uses SeedVR2.

UseModel~Size
T2I / img2img (default)AbstractFramework/z-image-turbo-8bit~11GB
T2I + reference edit (1–10 images)Qwen/Qwen-Image-2.1~33GB
Same · 8-step distillQwen/Qwen-Image-2.1-Turbo~13GB local
T2I / img2img only (no vision refs)Qwen-Image-2.1-Uncensored-MLXmanual local convert
Instruction edit / inpaint / outpaintmlx-community/Qwen-Image-Edit-8bit~37.5GB
Multi-reference (first image sets layout) / outpaintAbstractFramework/flux.2-klein-9b-8bit~18GB
Image upscaleAbstractFramework/seedvr2-3b-8bit~4.7GB
Vision caption for Agentgavinmroy/Qwen3-VL-8B-NSFW-Caption-V4.5-affine-4bit~5.4GB

Disk

Disk

h3.c (MiniMax-H3 alone is about 144GB) and the mlxgen image models are large, so allow tens to well over a hundred GB of free disk before the first run. The first-launch wizard does environment detection, component installation, and model downloads, and that step needs a connection.

You only need a connection to download models the first time and to update the app. During generation, video, images, and prompts are written into the local project and do not need to be uploaded.

Install

Install from the DMG

Four steps. If 0.1.0 is already on the Mac, quit it before you replace the app.

  1. Download the DMG

    Download the macOS Apple Silicon build from GitHub Releases: BigBotArt-0.1.1-arm64.dmg. The release notes give the first 24 hex characters of the SHA-256 as 67ba3c4947fecd6ed0cb380e.

  2. Drag it into Applications

    Double-click the DMG, drag BigBotArt into the Applications folder, and launch it from Applications. If 0.1.0 is already installed, quit the old app before replacing it.

  3. Finish the first-launch wizard

    On first launch, follow the wizard through environment detection, component installation, and model downloads. This step needs a connection, and the models are large, so allow tens to well over a hundred GB of free disk.

  4. Start the first project

    Describe the brief on Home, pick a Skill, and let the resident agent lay out the storyboard on the infinite canvas. Pick takes into the timeline and export with local FFmpeg.

The release notes page records only the v0.1.1 installer fix that was actually published. See v0.1.1 release notes. Contact jyllyj826@gmail.com.

Local files

Files stay on the Mac

The v0.1.1 release notes say projects, the canvas, takes, and assets are written only to ~/.bigbotart, assets are not uploaded, and there is no account and no subscription.

Source and releases live at github.com/jiangzhenguo/bigbot.art. The notes for this version are at v0.1.1.

Questions

Questions on this page

These answers match the visible text on this page and can be quoted as written.

  1. Which models and engines are used?

    The h3.c video engine runs MiniMax-H3 (BF16, ~144GB) for shot-level video with audio — text-to-video, first-last frame, and Ref2VA; optional style packs merge LoRA offline into a copy of the base model. mlxgen (package mlx-gen-jzg) downloads image models on demand: Z-Image-Turbo (default T2I/img2img), Qwen-Image-2.1 and Turbo (with references), Qwen-Image-Edit (instruction edit / inpaint / outpaint), FLUX.2 Klein 9B (multi-reference / outpaint), SeedVR2 3B (upscale), and a Qwen3-VL caption model for Agent vision; an Uncensored T2I variant needs a manual local convert. Agent chat uses any OpenAI-compatible endpoint (for example LM Studio). Final export runs through local FFmpeg; optional post-process Real-ESRGAN / GFPGAN / CodeFormer must be installed separately.

  2. Which Macs does BigBotArt v0.1.1 support?

    v0.1.1 ships only as a macOS Apple Silicon (arm64) build, and the release notes list M1, M2, M3, and M4. The exact minimum macOS version and memory requirements are listed on the GitHub Releases page.

  3. Where does BigBotArt store project files?

    The v0.1.1 release notes say projects, the canvas, takes, and assets are written only to the local directory ~/.bigbotart, and assets are not uploaded to a cloud.

  4. How do I check the BigBotArt 0.1.1 installer?

    The installer file is BigBotArt-0.1.1-arm64.dmg at https://github.com/jiangzhenguo/bigbot.art/releases/download/v0.1.1/BigBotArt-0.1.1-arm64.dmg . The release notes give the first 24 hex characters of the SHA-256 as 67ba3c4947fecd6ed0cb380e.