Files read on Hugging Face: 2026-10-31

FLUX 3 Image Klein on Mac: files, RAM and published times

Distilled 5.6B text-to-image and editing model with Qwen3-VL-4B-Instruct as the text encoder. The FLUX Kommunity License limits model use to non-commercial work, with an output exception below US$5,000,000 revenue.

Answer

FLUX 3 Image Klein runs on a Mac with 32 GB or more. A published run on M1 Max · 32 GB used the mflux 4-bit build (source, 2026-10-20).

Smallest Mac with a published run
32 GBM1 Max · 32 GB · mflux 4-bit
Smallest complete download
12.88 GBQ4_K_M GGUF
Files read
2026-10-31Hugging Face

FLUX Kommunity License v1.0. Commercial use: restricted. Excluded territories: none listed. Ordinary model use is non-commercial. Output generation has an exception for qualifying users below US$5,000,000 gross annualized revenue.

See the exact GGUF files and sizes

This is the distilled FLUX 3 Image Klein checkpoint. FLUX 3 Image Klein Base and the prompt upsampler are separate repositories and roles.

Table 1

What do you download?

Only builds with every required component and an exact probed file size are shown.

mflux 4-bit

safetensors · 4-bit (mflux --quantize 4) · 20.31 GB complete · Recommended

Mac routes: mflux

Exact component files for mflux 4-bit
FileComponentRequiredDownloadSource
transformer/diffusion_pytorch_model-00001-of-00002.safetensorsblack-forest-labs/flux-3-image-kleinFLUX 3 transformerditYes9.91 GBHugging FaceRead 2026-10-31
transformer/diffusion_pytorch_model-00002-of-00002.safetensorsblack-forest-labs/flux-3-image-kleinFLUX 3 transformerditYes1.19 GBHugging FaceRead 2026-10-31
text_encoder/model-00001-of-00002.safetensorsblack-forest-labs/flux-3-image-kleinQwen3-VL-4B-Instruct text encodertext_encoderYes4.97 GBHugging FaceRead 2026-10-31
text_encoder/model-00002-of-00002.safetensorsblack-forest-labs/flux-3-image-kleinQwen3-VL-4B-Instruct text encodertext_encoderYes3.91 GBHugging FaceRead 2026-10-31
vae/diffusion_pytorch_model.safetensorsblack-forest-labs/flux-3-image-kleinFLUX image autoencodervaeYes0.34 GBHugging FaceRead 2026-10-31

Build sources: mflux model registry · mflux README.

Q4_K_M GGUF

gguf · Q4_K_M · 12.88 GB complete

Mac routes: Diffusers

Exact component files for Q4_K_M GGUF
FileComponentRequiredDownloadSource
flux-3-image-klein-Q4_K_M.ggufunsloth/flux-3-image-klein-GGUFFLUX 3 transformerditYes3.67 GBHugging FaceRead 2026-10-31
text_encoder/model-00001-of-00002.safetensorsblack-forest-labs/flux-3-image-kleinQwen3-VL-4B-Instruct text encodertext_encoderYes4.97 GBHugging FaceRead 2026-10-31
text_encoder/model-00002-of-00002.safetensorsblack-forest-labs/flux-3-image-kleinQwen3-VL-4B-Instruct text encodertext_encoderYes3.91 GBHugging FaceRead 2026-10-31
vae/diffusion_pytorch_model.safetensorsblack-forest-labs/flux-3-image-kleinFLUX image autoencodervaeYes0.34 GBHugging FaceRead 2026-10-31

Build sources: FLUX 3 Image Klein GGUF card · Diffusers GGUF guide.

Q8_0 GGUF

gguf · Q8_0 · 15.30 GB complete

Mac routes: Diffusers

Exact component files for Q8_0 GGUF
FileComponentRequiredDownloadSource
flux-3-image-klein-Q8_0.ggufunsloth/flux-3-image-klein-GGUFFLUX 3 transformerditYes6.09 GBHugging FaceRead 2026-10-31
text_encoder/model-00001-of-00002.safetensorsblack-forest-labs/flux-3-image-kleinQwen3-VL-4B-Instruct text encodertext_encoderYes4.97 GBHugging FaceRead 2026-10-31
text_encoder/model-00002-of-00002.safetensorsblack-forest-labs/flux-3-image-kleinQwen3-VL-4B-Instruct text encodertext_encoderYes3.91 GBHugging FaceRead 2026-10-31
vae/diffusion_pytorch_model.safetensorsblack-forest-labs/flux-3-image-kleinFLUX image autoencodervaeYes0.34 GBHugging FaceRead 2026-10-31

Build sources: FLUX 3 Image Klein GGUF card · Diffusers GGUF guide.

BF16 safetensors

safetensors · BF16 · 20.31 GB complete

Mac routes: Diffusers

Exact component files for BF16 safetensors
FileComponentRequiredDownloadSource
transformer/diffusion_pytorch_model-00001-of-00002.safetensorsblack-forest-labs/flux-3-image-kleinFLUX 3 transformerditYes9.91 GBHugging FaceRead 2026-10-31
transformer/diffusion_pytorch_model-00002-of-00002.safetensorsblack-forest-labs/flux-3-image-kleinFLUX 3 transformerditYes1.19 GBHugging FaceRead 2026-10-31
text_encoder/model-00001-of-00002.safetensorsblack-forest-labs/flux-3-image-kleinQwen3-VL-4B-Instruct text encodertext_encoderYes4.97 GBHugging FaceRead 2026-10-31
text_encoder/model-00002-of-00002.safetensorsblack-forest-labs/flux-3-image-kleinQwen3-VL-4B-Instruct text encodertext_encoderYes3.91 GBHugging FaceRead 2026-10-31
vae/diffusion_pytorch_model.safetensorsblack-forest-labs/flux-3-image-kleinFLUX image autoencodervaeYes0.34 GBHugging FaceRead 2026-10-31

Build sources: FLUX 3 Image Klein model card.

Table 2

Does it fit your Mac's RAM?

Published names the exact measured machine. est. is a memory calculation. unknown means the run verdict is not established, not that the model fails.

Run verdicts by Mac memory tier
Mac RAMComplete buildDownloadRun verdictBasis
8 GBQ4_K_M GGUFQ4_K_M · Diffusers12.88 GBDoes not fitest.
16 GBQ4_K_M GGUFQ4_K_M · Diffusers12.88 GBunknownunknown
24 GBmflux 4-bit4-bit (mflux --quantize 4) · mflux20.31 GBunknownunknown
32 GBmflux 4-bit4-bit (mflux --quantize 4) · mflux20.31 GBRuns · PublishedPublishedApple M1 Max · 32 GB RAM
48 GBmflux 4-bit4-bit (mflux --quantize 4) · mflux20.31 GBRuns · Published on M1 Max · 32 GBPublishedApple M1 Max · 32 GB RAM
64 GBmflux 4-bit4-bit (mflux --quantize 4) · mflux20.31 GBRuns · Published on M1 Max · 32 GBPublishedApple M1 Max · 32 GB RAM
96 GBmflux 4-bit4-bit (mflux --quantize 4) · mflux20.31 GBRuns · Published on M1 Max · 32 GBPublishedApple M1 Max · 32 GB RAM
128 GBmflux 4-bit4-bit (mflux --quantize 4) · mflux20.31 GBRuns · Published on M1 Max · 32 GBPublishedApple M1 Max · 32 GB RAM

Every image model, ranked by memory: 8 GB Mac · 16 GB Mac · 24 GB Mac · 32 GB Mac · 64 GB Mac

Table 3

Does it fit your NVIDIA GPU's VRAM?

Runs · Stated means the linked model card, app maker or NVIDIA says a GPU with that much VRAM runs the model; Runs · Published names the card of a dated run. Fits with offload means the main model weights fit in VRAM with a 15% margin but the full set does not: the text encoder and VAE run from system RAM, which is slower. Exceeds VRAM means even the main weights are too large for the card. Mac runs never count here.

Run verdicts by NVIDIA VRAM tier
NVIDIA VRAMComplete buildDownloadRun verdictBasis
8 GBNot build-specificunknownRuns · StatedStated8 GB · official ComfyUI int8 files · FLUX 3 Image Klein README (Black Forest Labs)
12 GBNot build-specificunknownRuns · StatedStated8 GB · official ComfyUI int8 files · FLUX 3 Image Klein README (Black Forest Labs)
16 GBNot build-specificunknownRuns · StatedStated8 GB · official ComfyUI int8 files · FLUX 3 Image Klein README (Black Forest Labs)
24 GBNot build-specificunknownRuns · StatedStated8 GB · official ComfyUI int8 files · FLUX 3 Image Klein README (Black Forest Labs)
32 GBNot build-specificunknownRuns · StatedStated8 GB · official ComfyUI int8 files · FLUX 3 Image Klein README (Black Forest Labs)

Every image model, ranked by memory: 8 GB of VRAM · 12 GB of VRAM · 16 GB of VRAM · 24 GB of VRAM

How does it rank in blind votes?

No independent score yet.

Table 4

What published Mac times are available?

Third-party results keep their original machine and settings. ModelFit does not run these benchmarks.

Published Mac generation measurements
Machine and RAMSettingsAppResultSource
Apple M1 Max · 32 GB RAM512x512 · 16 steps · 4-bit (mflux --quantize 4) · Default mode. Seed 42, mflux 0.23.0, MLX 0.32.0, macOS 14.6, local 4-bit conversion of the distilled checkpoint. External wall time and peak MLX allocation. Other applications were active.mfluxSuccess49.5 s6.7 GB peakmflux FLUX 3 memory benchmarkPublished 2026-10-20 · Grade A
Apple M1 Max · 32 GB RAM512x512 · 16 steps · 4-bit (mflux --quantize 4) · --low-ram mode. Seed 42, mflux 0.23.0, MLX 0.32.0, macOS 14.6, local 4-bit conversion of the distilled checkpoint. External wall time and peak MLX allocation. Other applications were active.mfluxSuccess52.6 s4 GB peakmflux FLUX 3 memory benchmarkPublished 2026-10-20 · Grade A

Which Mac software routes are listed?

  • Diffusers

    Official

    Formats: safetensors, gguf. Memory policy: component stages stay resident.

    Runs on the PyTorch mps device on Mac and CUDA on NVIDIA. GGUF loads through from_single_file on the model class, not the pipeline. Resident size is the sum of the weights, an upper bound.

    Sources: Diffusers MPS guide · Diffusers GGUF guide

  • mflux

    Community · 0.23.0

    Formats: safetensors. Memory policy: component stages stay resident.

    MLX-native runner. It reads the original Hugging Face weights and quantizes them while loading (--quantize 3, 4, 5, 6 or 8). --low-ram lowers memory use at a performance cost.

    Install with uv

    uv tool install --upgrade mflux

    Sources: mflux README, install · mflux pyproject · mflux common README, quantization

  • stable-diffusion.cpp

    Official

    Formats: gguf, safetensors. Memory policy: component stages stay resident.

    C/C++ runner with Metal on Mac and CUDA on NVIDIA. It reads GGUF and safetensors, and its docs keep a page per model family. Resident size is the sum of the weights, an upper bound.

    Sources: stable-diffusion.cpp README, backends

What does the official configuration say?

"Peak GPU memory with the reference code on a B200 at 1024x1024 with the prompt upsampler off is 31 GB in BF16, 26 GB in FP8 and 25 GB in NVFP4. This is a datacenter GPU figure, not a Mac minimum."

Source: FLUX 3 Image Klein model card · Read 2026-10-24.

What known issues should you check?

  • Known issue 1

    Black Forest Labs says not to use the model without a prompt upsampler. The upsampler is a separate 9B model and is not counted in the sizes here.

    Source: FLUX 3 Image Klein model card · Read 2026-10-24

  • Known issue 2

    The Mac timings come from a run with other applications active and growing swap. The authors call them observations under shared load, not a speed ranking.

    Source: mflux FLUX 3 memory benchmark · Read 2026-10-24

Frequently asked questions

How fast is FLUX 3 Image Klein on a Mac?

An mflux 4-bit conversion on an M1 Max with 32 GB made a 512x512 image in 16 steps in 49.45 s in default mode, peaking at 6.71 GB of MLX memory. Low-RAM mode took 52.63 s with a 3.997 GB peak. Other applications were running.

Do I need the FLUX 3 prompt upsampler?

Black Forest Labs says not to use the model without a prompt upsampler. It is a separate 9B model and is not counted in the file sizes here, so those sizes understate the official workflow.

Can I use FLUX 3 Image Klein commercially?

Ordinary model use is non-commercial under the FLUX Kommunity License v1.0. Output generation has an exception for qualifying users below US$5,000,000 gross annualized revenue. Read the license text for the full list of prohibited uses.

Sources