FLUX 3 Image Klein: Is It Free, and How Do You Use It?

2026-10-17 · 5 min read

FLUX 3 Image Kleinopen weightsfree accessmodel-release

Black Forest Labs released FLUX 3 Image Klein on October 15, 2026: a 5.6B open-weights model for text-to-image generation and image editing that takes a prompt and up to 10 reference images and returns one image. The weights are free to download under the FLUX Kommunity License, but there is no free hosted API — the BFL API and OpenRouter both bill per image. aiwith has not integrated FLUX 3 Image Klein; if you want to generate an image on aiwith today, use its built-in image generator, which runs Nano Banana Pro, Seedream 5.0, GPT Image 2.5 and Qwen Image.

Last checked: October 17, 2026. This is a source-checked access guide, not a hands-on benchmark.

What is FLUX 3 Image Klein?

According to Black Forest Labs' release post and the Hugging Face model card, the documented specifications are:

5.6B parameters for the image transformer (the text encoder is counted separately).
Text-to-image, single-reference editing and multi-reference generation in one model, with up to 10 reference images in a request.
Native 4K generation (generated at 4K, not upscaled into it), pixel-perfect editing, and bounding boxes placed on a 0–1000 grid in the prompt. Boxes guide placement and scale; they are not a hard pixel mask.
Text rendering inside generated images, and editing of text that is already there.
The text encoder is Qwen3-VL-4B-Instruct. A separate 9B prompt upsampler — a fine-tune of Qwen3.5-9B released under Apache 2.0 — expands a short request into the full scene caption the model expects. BFL warns that a short prompt sent as typed gives generic output, so plan for the upsampler.
Two checkpoints: Klein Base (undistilled, bfloat16, 50 steps, guidance 3.5) for fine-tuning, LoRA training and research, and Klein (distilled, bfloat16, 16 steps) for generation and editing — the checkpoint behind the BFL API. Both ship alongside FP8, INT8 and NVFP4 builds made with NVIDIA support.

In BFL's own internal evaluations, Klein ranks first among open-weights models in text-to-image, multi-reference generation and image editing, and above the closed models in its comparison (Ideogram V4.5 and FLUX.2 [pro]). Those are vendor numbers — treat them as claims, not an independent benchmark.

Is FLUX 3 Image Klein free?

There are three different questions behind “free”:

Are the weights free? Yes. The checkpoints are downloadable on Hugging Face under the FLUX Kommunity License. BFL's post describes the license as allowing “free commercial use for individuals and companies with annual revenue under $5M, alongside research and non-commercial use,” with larger organizations needing a paid self-hosting license or the API. Read the license text yourself before building a product on it — “open weights” is not the same as an open-source license.
Is there a free hosted option? Not from BFL. BFL does not publish a free tier or free trial credits, and its Playground is priced the same as the API. A free community demo does exist on Hugging Face Spaces, and ComfyUI has day-one support — but those are demos and tooling, not a free production API. At this check OpenRouter listed no free variant of the model.
Is the API free? No. BFL and OpenRouter both bill per image.

How do you use FLUX 3 Image Klein?

The route depends on what you need:

Fastest free look — Hugging Face Space. The official Space runs text-to-image, editing and bounding boxes in the browser with no setup. It is a demo, so expect queueing and limits.
Self-host the weights. Download the checkpoint from Hugging Face and run it through ComfyUI or Hugging Face Diffusers, both of which have day-one support, or the reference inference code on GitHub. This is the genuinely free route, and it needs the hardware covered below.
Paid API. BFL serves the distilled checkpoint as flux-3-image-klein; OpenRouter lists it as black-forest-labs/flux-3-image-klein. Each provider has its own account and credit rules.

What does FLUX 3 Image Klein cost?

BFL API: pricing starts at $0.009 per 1K image, per BFL's release post.
OpenRouter: from $0.008/image at 768 up to $0.036/image at 4K, according to the OpenRouter model page we read on October 17, 2026.

Prices and provider availability change, so check the provider's own page before you rely on a number copied here.

Can you run FLUX 3 Image Klein on your own hardware?

Yes — this is the part that changed from the larger FLUX 3 Image model. Per BFL's release post:

The generator weights need 11.2 GB in BF16 and 5.6 GB in FP8.
With the INT8 build, Klein runs on NVIDIA GeForce RTX GPUs with as little as 8 GB of VRAM, through ComfyUI or Diffusers.
With the distilled model and the INT8 build, a 1MP image generates in about 4 seconds on an RTX 5090 and under 9 seconds on an RTX 4070; on an NVIDIA GB300 the distilled model makes a 1MP image in under 0.6 seconds in NVFP4.

Remember that the prompt upsampler is a separate ~18.8 GB download. The reference CLI loads it by default; --no-pup sends the prompt as typed, which is only a good idea if your prompt is already a full scene caption.

Is FLUX 3 Image Klein available on aiwith?

No, as of this check. An aiwith subscription does not unlock FLUX 3 Image Klein, and it is not in our model line-up.

If you mainly want to generate an image in a browser, open the AI image generator on aiwith and pick one of the models it actually runs:

Nano Banana Pro — the strongest overall of the in-house image options.
Seedream 5.0 — a balance between quality and cost.
GPT Image 2.5 — sharper detail and stronger editing.
Qwen Image — stronger text rendering in images (logos, posters, UI mockups).

These are different models and will not match FLUX 3 Image Klein's output — treat them as the practical way to make an image on aiwith, not as access to FLUX 3. If your goal is specifically to evaluate Klein, use the Hugging Face Space or a provider that explicitly lists the model.