Skip to content
English

Switch theme

// FLUX · Image generation

FLUX.1-Krea-dev: VRAM & local deployment

11.9B parameters. About 8.4 GB at 4-bit and 26.7 GB at BF16.

released 2025-07-07licence: Non-commercialmemory figures are estimates

01Specs

PublisherBlack Forest Labs · black-forest-labs
Family / seriesFLUX · FLUX.1-Krea-dev
TypeImage generation (text-to-image)
Parameters11.9B Source: counted from official safetensors weights
Released weightsBF16 · 24.1 GBdownload
Licenceflux-1-dev-non-commercial-license · Non-commercialResearch / non-commercial licence: commercial use is not allowed without a separate agreement.
Released2025-07-07 (Hugging Face repo created)
AccessGated: accept the licence on Hugging Face before downloading.

02How much VRAM it needs (estimate)

Image models: memory depends on resolution, batch and which text encoders are kept on the GPU; Diffusers / ComfyUI can offload parts to CPU. The weight size below is the main transformer only.

PrecisionWeightsWeights + overheadSmallest GPUMac
BF16 / FP16 (original)23.8 GB 26.7 GB RTX 5090 32GBMac 64GB
FP8 / INT8 / Q8_012.6 GB 14.4 GB 16GB GPUMac 24GB
4-bit (Q4_K_M / AWQ / GPTQ-Int4)7.2 GB 8.4 GB 12GB GPUMac 16GB

Bits per weight: BF16 16, FP8 8.5, 4-bit 4.85. “Fits” = total ≤ 90% of memory; Mac GPU can use ~2/3 of unified memory (≤36GB) or ~3/4 (larger). For MoE, all experts must be in memory, so size by total parameters. Method →

03How to run it

Only runtimes the publisher or the runtime’s official library names for this model.

Which inference engine to choose →

04Official repositories & quantised builds

1 official repo(s) from black-forest-labs.

05Other FLUX models

All 14 →
ModelParams4-bit totalReleased
flux-3-image-klein-pup9.41B6.8 GB2026-10-14
flux-3-image-klein5.55B4.2 GB2026-10-13
FLUX.2-klein-9b-kv9.08B6.6 GB2026-03-09
FLUX.2-klein-4B3.88B3.1 GB2026-01-14
FLUX.2-klein-9B9.08B6.6 GB2026-01-14
FLUX.2-dev32.2B22.0 GB2025-11-22
FLUX.1-Kontext-dev11.9B8.4 GB2025-05-28
FLUX.1-Canny-dev11.9B8.4 GB2024-11-20

06Sources

Data checked 2025-07-31. Parameters, context and licence are copied from the official repository; memory figures are our estimates.

Data updated: Self-host open models(UTC+8)