MiniMax H3 Pruned BF16: Compatible GPUs, VRAM and Download

Main model (diffusion) · BF16 · 40.23GB · pruned · updated 2026-09-28

Item	Value
File name	`minimax_h3_fl2va_pruned_bf16.safetensors`
Component	Main model (diffusion)
Quant / format	BF16
Size	40.23 GB
ComfyUI folder	diffusion_models (main model folder)
Source	Comfy-Org/MiniMax-H3

Which GPUs can run it

736 × 416 · 5s: 30 GPUs workable

GTX 1080 Ti 11GB RTX 2060 12GB RTX 2080 Ti 11GB RTX 2080 Ti 22GB (modded) RTX 3060 12GB RTX 3070 16GB (modded) RTX 3080 10GB RTX 3080 12GB RTX 3080 20GB (modded) RTX 3080 Ti 12GB RTX 3090 24GB RTX 4060 Ti 16GB RTX 4070 12GB RTX 4070 Super 12GB RTX 4070 Ti 12GB RTX 4070 Ti Super 16GB RTX 4080 16GB RTX 4080 32GB (modded) +12 more

1344 × 768 · 5s: 19 GPUs workable

RTX 2080 Ti 22GB (modded) RTX 3070 16GB (modded) RTX 3080 20GB (modded) RTX 3090 24GB RTX 4060 Ti 16GB RTX 4070 Ti Super 16GB RTX 4080 16GB RTX 4080 32GB (modded) RTX 4080 Super 16GB RTX 4080 Super 32GB (modded) RTX 4090 24GB RTX 4090 D 24GB RTX 4090 48GB (modded) RTX 5060 Ti 16GB RTX 5070 Ti 16GB RTX 5080 16GB RTX 5090 32GB RTX 5090 D 32GB +1 more

1344 × 768 · 15s: 19 GPUs workable

“Workable” = estimated status ok or warn with 32GB RAM / fast-disk (warn means weight offloading and noticeably slower).

Every quant in this family, side by side

Same family = different quants of the same base variant. They are interchangeable — what changes is size, speed and quality loss. "Peak" is a rough VRAM estimate (file size × 0.6, floor 8G).

File	Quant	Size	Pruned	VRAM fit
P- W4A8	W4A8	12.54 GB (peak ~8G)	pruned	for 12-16GB cards
P- FP8	FP8 Scaled	20.96 GB (peak ~13G)	pruned	wants 24GB+
P- INT8	INT8 ConvRot	20.97 GB (peak ~13G)	pruned	wants 24GB+
INT8	INT8 ConvRot	34.04 GB (peak ~20G)	full	wants 24GB+
P- BF16 ← this page	BF16	40.23 GB (peak ~24G)	pruned	wants 24GB+
BF16	BF16	66.28 GB (peak ~40G)	full	wants 24GB+

Rule of thumb: anything under 10GB fits an 8GB card, the 10-16GB tier targets 12-16GB cards, and files above 16GB want 24GB+. Within this family the flagged P- INT8 is the official default — take it unless you have a reason not to. The smallest here is P- W4A8 (12.54GB) and the largest is BF16 (66.28GB); this file (40.23GB) sits in the "wants 24GB+" tier. About this quant: the full-precision baseline: highest quality ceiling, biggest files — mostly for large-VRAM, max-quality setups. "Pruned" trims redundant channels for size — the quality cost is usually tiny.

Who is this file for?

Role in the pipeline: the main model — it sets generation quality and speed and is the biggest download; swapping quants changes detail and pace, not how you write prompts.Frankly, a 40.23GB file is all but guaranteed to run through weight offloading unless you have 40GB+ of VRAM. Most people should grab a 10-25GB pruned INT8 or W4A8 build from the table above instead (this family's P- INT8 at 20.97GB and P- W4A8 at 12.54GB are a fraction of the size); this one is for "VRAM to spare, full precision only" setups. Full-precision weights also read from disk more slowly, so expect a longer first load.

Fastest vs. minimum hardware

"Fastest" = the card with the shortest estimated time; "floor" = the smallest-VRAM card that still runs it. Times include model load and text encoding; the first run also compiles kernels.

736 × 416 · 5s: fastest RTX 5090 32GB at ~1 min 15 s (30/49 GPUs workable) · floor RTX 3080 10GB (needs weight offloading — noticeably slower)
1344 × 768 · 5s: fastest RTX 5090 32GB at ~3 min 17 s (19/49 GPUs workable) · floor RTX 4060 Ti 16GB (needs weight offloading — noticeably slower)
1344 × 768 · 15s: fastest RTX 5090 32GB at ~20 min 4 s (19/49 GPUs workable) · floor RTX 4060 Ti 16GB (needs weight offloading — noticeably slower)

Download, place, refresh — 3 steps

Grab it: the button above redirects to the official HuggingFace / GitHub repo — /go is just a redirect, we never rehost. Skip cloud-drive mirrors; if HuggingFace is slow or blocked on your network, a mirror can speed up the transfer, but verify the file against the official repo. Keep at least 2× the file size free on disk.
Place it: drop the file into diffusion_models (main model folder) exactly as shown in the spec table, filename untouched — loaders locate files by folder and name, so a rename makes it "vanish".
Load it: hit Refresh in ComfyUI (it only scans folders at startup or on refresh), then pick the file in the loader node. loader node: Load Diffusion Model (GGUF files go through the ComfyUI-GGUF Unet Loader).

Sanity check: once downloaded, compare the size on disk with the 40.23GB listed here. Way off? The download is incomplete — browsers sometimes keep a half-written file looking like a finished one — so delete it and pull again.

Also required: It takes all three to render: besides this file you need the text encoder NVFP4 AWQ and the VAEs BF16 + BF16 — miss any one of them and the workflow errors out on the spot.

Won't load? Check these three

Nothing shows up after Refresh: nine times out of ten it's the wrong folder — re-check the ComfyUI folder row in the spec table above. The file must sit inside ComfyUI's models/ subfolder, and no, a desktop shortcut doesn't count.
Instant OOM the moment it loads: make sure no other big model is sitting in VRAM first, then step down to a smaller quant from the table above — the INT8 vs BF16 guide has the full math on why brute-forcing backfires.
Renamed it and now nothing picks it up: loaders match by filename (GGUF metadata stores the original name too), so restore the original name and hit Refresh once more.

Download · HuggingFace →

→ Run the numbers for your RAM/budget in the calculator (this file pre-filled)

FAQ

How much VRAM does minimax_h3_fl2va_pruned_bf16.safetensors need?

The file is 40.23GB (Main model (diffusion)). Runtime VRAM also depends on resolution and duration; a 720p-class short clip typically needs roughly 24GB+ of free VRAM. Run the calculator for your exact card.

Where does this file go in ComfyUI?

diffusion_models (main model folder). Keep the filename as-is; refresh ComfyUI and select it in the matching loader.

How does it compare with other quantizations?

See sibling entries in the model library and the INT8/BF16 comparison article. Rule of thumb: 8GB → GGUF Q3/Q2; 12-16GB → pruned INT8 or W4A8; 24GB+ → full INT8 or NVFP4 (RTX 50).

Similar files: BF16 · INT8 · P- INT8 · P- FP8 · BF16 · INT8 · P- BF16 · P- INT8 · P- FP8 · P- W4A8