Black Forest Labs' FLUX.2 Klein 4B Hits 900K Downloads With ComfyUI Support
Black Forest Labs' smallest Flux.2 model lands in ComfyUI as a single 7.75GB file, generating images in under a second on an RTX 3090.
- Comfy-Org repackaged Flux.2 Klein 4B as a single 7.75GB safetensors for ComfyUI, 900K+ downloads.
- 4B rectified flow transformer generating or editing images in under 0.5 seconds on modern GPUs.
- Fits in ~13GB VRAM, runs on RTX 3090 and RTX 4070 class hardware.
- Apache 2.0 license on the 4B weights allows unrestricted commercial use.
- Step and guidance distilled to just 4 inference steps from a larger teacher model.
- Hosted API priced at $0.014 per megapixel, with LoRA fine-tuning now in public beta.
Black Forest Labs’ speed-focused FLUX.2 [klein] 4B is now available in a ComfyUI-oriented package. The ComfyUI model page provides the transformer weights as one safetensors file, sparing developers from mapping the original Diffusers directory structure to ComfyUI loaders.
The 4-billion-parameter model combines text-to-image generation and multi-reference editing in a compact rectified-flow architecture. Black Forest Labs designed the FLUX.2 Klein family for low-latency inference, and the 4B release uses Apache 2.0 terms that permit commercial deployment.
At publication, the Hugging Face repository had recorded more than 900,000 downloads. Repeated and automated pulls contribute to that counter, but the volume still indicates substantial interest from local ComfyUI users.
One file, plus the required dependencies
| Property | Details |
|---|---|
| Architecture | 4-billion-parameter rectified-flow transformer |
| Tasks | Text-to-image generation and multi-reference editing |
| Sampling | Four inference steps |
| Model file | One 7.75 GB safetensors file |
| Reported VRAM | About 13 GB, depending on workflow and precision |
| License | Apache 2.0 for the 4B weights |
The safetensors file contains the central transformer weights. Matching text-encoder, VAE, and workflow assets remain separate downloads, so the package simplifies the largest part of the installation without collapsing the entire pipeline into one file.
From download to first render
- Update ComfyUI and its frontend to a version with FLUX.2 [klein] support.
- Download the 7.75 GB safetensors file from the model repository.
- Copy the file into
ComfyUI/models/diffusion_models. - Download the text-encoder and VAE files referenced by the repository’s example workflow, then place them in their corresponding ComfyUI model directories.
- Restart ComfyUI or refresh its model list, load the example workflow, and select the downloaded files in the relevant loader nodes.
This story is for Pro members
You've reached the end of the free preview. Upgrade to AlphaSignal Pro to read the full article - and everything else behind the paywall.
Comments
