FLUX 3 Image Klein
Needs 7 GB at FP8 / INT8. The smallest real configuration today is 1× RTX 3060 at Vast.ai, $0.04 per hour, about $26 a month at 730 hours.
- Per hour
- $0.04
- Per month
- $26
- Cheapest at
Vast.ai
What is documented about FLUX 3 Image Klein
news checked 2026-10-27Step-distilled (16 steps, no guidance). Text-to-image, editing and multi-reference (up to 10 images), bounding-box layout, native 4K. Files: transformer 11.10 GB BF16 (HF metadata 5.55B params), text encoder Qwen3-VL-4B-Instruct 8.88 GB, FLUX.2 VAE 0.34 GB; fp8r variant 6.27 GB, nvfp4 5.28 GB. Card says it is not meant to run without a prompt upsampler (see FLUX 3 Image Klein Prompt Upsampler). Speed (BFL): INT8 1 MP in ComfyUI about 4 s on RTX 5090, under 9 s on RTX 4070; NVFP4 under 0.6 s on GB300. Also covers image editing (single checkpoint). HF repo created 2026-10-13, public launch 2026-10-15.
- https://huggingface.co/black-forest-labs/flux-3-image-klein
- https://huggingface.co/api/models/black-forest-labs/flux-3-image-klein?blobs=true
- https://bfl.ai/blog/flux-3-image-klein
How much VRAM it needs
Weights plus 20% runtime overhead. For image generation models, activations also depend on resolution, clip length and batch size; the study's per-GPU estimates below give the practical configuration.
| Precision | Weights | Weights + headroom |
|---|---|---|
| BF16published checkpoint | 43 GB | 52 GB |
| FP8 / INT8estimated | 6 GB | 7 GB |
| INT4 / MXFP4estimated | 3 GB | 5 GB |
What it costs to run, GPU by GPU
The fewest GPUs of one type whose total memory fits the model at short context, priced from the cheapest eligible on-demand offer for exactly that count (45-day window, newest reading per provider), with the next providers beside it. Monthly = 730 hours.
| GPU | Smallest config | VRAM used | $ / hour | $ / month | Cheapest at |
|---|---|---|---|---|---|
| RTX 3060Best fit | 1× 12 GB12 GB total | 56% | $0.04 | $26 | |
| RTX A2000 | 1× 12 GB12 GB total | 56% | $0.04 | $26 | |
| Tesla P4 | 1× 8 GB8 GB total | 84% | $0.04 | $29 | |
| T4 | 1× 16 GB16 GB total | 42% | $0.04 | $29 | |
| Tesla P40 | 1× 24 GB24 GB total | 28% | $0.04 | $30 | |
| GTX Titan X | 1× 12 GB12 GB total | 56% | $0.04 | $31 | |
| Titan Xp | 1× 12 GB12 GB total | 56% | $0.04 | $31 | |
| GTX 1080 | 1× 8 GB8 GB total | 84% | $0.04 | $32 | |
| Tesla P100 | 1× 16 GB16 GB total | 42% | $0.04 | $33 | |
| GTX 1070 | 1× 8 GB8 GB total | 84% | $0.05 | $36 | |
| RTX 4060 | 1× 8 GB8 GB total | 84% | $0.05 | $40 | |
| RTX 2070 | 1× 8 GB8 GB total | 84% | $0.06 | $41 | |
| RTX 3060 Ti | 1× 8 GB8 GB total | 84% | $0.06 | $41 | |
| RTX 2060 Super | 1× 8 GB8 GB total | 84% | $0.06 | $41 | |
| Quadro P2000 | 2× 5 GB10 GB total | 67% | $0.06 | $41 | |
| GTX 1070 Ti | 1× 8 GB8 GB total | 84% | $0.06 | $44 | |
| RTX 2070 Super | 1× 8 GB8 GB total | 84% | $0.06 | $44 | |
| RTX 3070 | 1× 8 GB8 GB total | 84% | $0.06 | $44 | |
| GTX 1080 Ti | 1× 11 GB11 GB total | 61% | $0.06 | $45 | |
| RTX A4000 | 1× 16 GB16 GB total | 42% | $0.06 | $46 | |
| Tesla V100 16GB | 1× 16 GB16 GB total | 42% | $0.06 | $46 | |
| RTX 2080 | 1× 8 GB8 GB total | 84% | $0.06 | $47 | |
| RTX 2080 Super | 1× 8 GB8 GB total | 84% | $0.06 | $47 | |
| RTX 2080 Ti | 1× 11 GB11 GB total | 61% | $0.07 | $49 | |
| The 24 cheapest of 104 GPU types that fit. Every GPU, any precision and context, in the finder | |||||
