Datasheet for local AI49 models98 GPUsData read 2026-09-25
No adsNo tracking
Check my GPU →

Can the Mac 48 GB run MiniMax H3 Pruned?

Video model · pruned 33B48 GB unified · ~40.2 GB for the GPU · 273–614 GB/s · Apple SiliconData 2026-09-25
Runs well
Yes — and comfortably.

Download Q8_0 (21.6 GB). With the model's working memory it needs about 27.4 GB, leaving 12.8 GB spare on 40.2 GB. Quality: practically identical to the original.

Best fileQ8_0
File size21.6 GB
GPU memory needed~27.4 GB
Shared memory85 / 48 GB
Memory map · Mac 48 GB27.4 GB / 40.2 GB
020 GB40 GB
Weights Q8_0 · 21.6 GBWorking memory · 5.0 GBReserve · 0.8 GBFree · 12.8 GB
calculated from real file sizes plus working memory. Not yet measured on this setup. How this works.

01What to download

PartFileFolderSize
ModelMiniMax-H3-FL2VA-Pruned-Q8_0.gguf
Q8_0 · Abiray/MiniMax-H3-Pruned-GGUF
models/unet21.6 GBDownload →
Text encoderqwen3vl_32b_minimax_h3_bf16.safetensors
Qwen3-VL 32B BF16 · Comfy-Org/MiniMax-H3
models/text_encoders51.5 GBDownload →
VAEminimax_h3_video_vae_fp16.safetensors
MiniMax H3 video VAE FP16 · Comfy-Org/MiniMax-H3
models/vae5.2 GBDownload →
Also neededminimax_h3_audio_vae_fp32.safetensors
MiniMax H3 audio VAE FP32 · Comfy-Org/MiniMax-H3
models/vae0.6 GBDownload →
Total download · keep about the same free on disk78.9 GB

On a Mac, FP8 text encoders do not load, so a 16-bit one is listed. Sizes read from Hugging Face (2026-09-25). “Download” links start the file directly; the file name links are the same files the official ComfyUI workflows use.

Shared memory: about 84.9 GB of your 48 GB. On this machine the model, the text encoder, the VAE and the operating system all use the same memory. That is about 36.9 GB more than there is. It still runs: while the model works, the system compresses or swaps the idle text encoder, so loading and changing the prompt get slower. A smaller file or text encoder avoids that. calculated

02Every MiniMax H3 Pruned file on 40.2 GB

FileSizeNeededOn this cardQualityDownload
16-bit
SAFETENSORS · Comfy-Org
40.2 GB46.0 GB5.8 GB too bigthe original weightsHugging Face →
Q8_0 ←
GGUF · Abiray
21.6 GB27.4 GBfits · 12.8 GB sparepractically identical to the originalHugging Face →
Q6_K
GGUF · Abiray
16.7 GB22.5 GBfits · 17.7 GB sparevery close to the originalHugging Face →
Q5_K_M
GGUF · Abiray
14.1 GB19.9 GBfits · 20.3 GB spareclose; small differences in fine detailHugging Face →
Q4_K_M
GGUF · Abiray
11.6 GB17.4 GBfits · 22.8 GB sparegood; some loss in fine detail and textHugging Face →
Q3_K_M
GGUF · Abiray
8.9 GB14.7 GBfits · 25.5 GB sparenoticeable loss of detailHugging Face →

Sizes from the Hugging Face file listing, read 2026-09-25. “Needed” = file + 5 GB working memory for this model + 0.8 GB kept free for the system. FP8 and INT8 files are left out on a Mac: Apple GPUs cannot compute FP8, so ComfyUI either fails to load them or converts them back to 16-bit, which saves no memory. Use a 16-bit or GGUF file.

03The text encoder

Qwen3-VL 32B: 51.5 GB as 16-bit, 27.1 GB as INT8, 15.7 GB as NVFP4. ComfyUI encodes the prompt first and can push the encoder out of VRAM before sampling, so it does not have to fit together with the model. On 40.2 GB the INT8 encoder fits on its own, so prompt encoding stays fast.

04About this GPU

A Mac with 48 GB shares that memory between the CPU and the GPU. On current macOS the GPU may use about 40.2 GB of it by default (older macOS versions: about 38.7 GB); that is the figure used here. ComfyUI can go past it, but macOS then starts compressing and swapping memory and everything slows down. ComfyUI runs on Apple GPUs through PyTorch's MPS backend. 16-bit and GGUF files work; FP8 and INT8 files do not save memory on a Mac, so they are skipped. Speed is the catch: even the fastest Macs are several times slower per image than a desktop RTX card. Chips sold with 48 GB: M3 Max, M4 Pro, M4 Max, M5 Pro, M5 Max (memory bandwidth 273–614 GB/s — the higher, the faster). Everything about Macs and local AI →

Mac 48 GB: 48 GB unified · ~40.2 GB for the GPU · 273–614 GB/s · Apple Silicon. Everything that runs on the Mac 48 GB →

05What more VRAM would change

Nothing to gain for this model: the Mac 48 GB already runs a top-quality file entirely in VRAM. More memory would only help with bigger images, longer clips or several models at once.

06Measured and reported results

Nobody has sent measured numbers for this pair yet. If you run MiniMax H3 Pruned on a Mac 48 GB, send your time per image and peak VRAM and it will appear here, credited.

07Questions

How much VRAM does MiniMax H3 Pruned need?

Around 14.7 GB with the smallest file (Q3_K_M) and 26.8 GB with an 8-bit file (FP8), counting working memory and a small system reserve. The full 16-bit file needs about 46.0 GB.

Which MiniMax H3 Pruned file should I download for the Mac 48 GB?

Q8_0 (21.6 GB) from Abiray/MiniMax-H3-Pruned-GGUF. Load it with Unet Loader (GGUF) from the ComfyUI-GGUF node pack.

Is FP8 faster than GGUF on the Mac 48 GB?

No. This GPU has no FP8 compute, so ComfyUI converts FP8 weights back before the maths. FP8 only saves memory here; Q8_0 GGUF is the closer-to-original 8-bit pick.

How much do I need to download for MiniMax H3 Pruned on the Mac 48 GB?

About 78.9 GB for the model file, text encoder and VAE listed on this page, and the same again free on disk. With a 40.2 GB card, 96 GB of system RAM or more is recommended.