Black Forest Labs: FLUX.2 [klein] 9B KV

FLUX.2 [klein] 9B KV is a KV-cache optimized variant of the Klein 9B model that caches reference image key-value pairs after the first denoising step, skipping redundant computation on subsequent steps. This delivers up to 2.5x faster inference for multi-reference editing tasks while retaining all capabilities of the standard Klein 9B, including sub-second text-to-image and advanced editing in 4 steps.

Compare models

flux-2-klein-9b-kv

AvailableBlack Forest LabsImageSync

Price: From $0.00078
Modalities: Image

About FLUX.2 [klein] 9B KV

FLUX.2 [klein] 9B KV is a variant of Black Forest Labs' 9-billion-parameter klein model that caches the key-value pairs of reference images after the first denoising step and reuses them on later steps. Black Forest Labs reports up to 2.5 times faster inference on multi-reference editing, while it keeps the text-to-image and editing abilities of the standard 9B.

Where it works well

Reference-image key-value pairs are computed once and reused, so the repeated work of re-reading reference images disappears after step one.
Black Forest Labs reports up to 2.5 times faster inference for multi-reference editing than the standard klein 9B.
It keeps the four-step, sub-second text-to-image behavior of the standard klein 9B.
One model covers generation and editing with several reference images.

When to choose another model

The speed gain applies to editing with reference images; plain text-to-image has nothing to cache and runs like the standard 9B.
Text in images may be rendered inaccurately, and the maker notes prompt adherence varies with phrasing.
It is released under a non-commercial license, so self-hosting it in a product needs a separate agreement.

Getting started

Create API key
Create a key in Console, then use it with every model on the platform.

Send your first request

Copy the example for your language and run it against the endpoint.

Image editingText to imageTokenLab endpoint

POST/v1/images/generations

curl -X POST "https://api.tokenlab.sh/v1/images/generations" \
  -H "Authorization: Bearer sk-xxx" \
  -H "Content-Type: application/json" \
  -d '{
  "operation": "image-to-image",
  "model": "flux-2-klein-9b-kv",
  "prompt": "A minimalist product photo of matte black headphones on a soft blue background.",
  "image_url": "https://example.com/source.png"
}'

Pricing

Official price is the model maker's public baseline. TokenLab price is what you pay for this model on TokenLab.

Image editing

per request

Official price: $0.00078
Official: $0.00078
Discount: —

Text to image

per request

Official price: $0.00078
Official: $0.00078
Discount: —

	Official priceper request	Officialper request	Discount
Image editing	$0.00078	$0.00078	—
Text to image	$0.00078	$0.00078	—

Usage & activity

Success rate is the share of requests that completed. Latency is how long a full response takes; P95 means 95% of requests finished within that time.

Usage & availability

Last 24 hours

Requests
Success rate
P95 latency
Total tokens

Model performance

Last updated Oct 3, 2026, 07:26 PM UTC

Metrics appear once privacy and data volume thresholds are met.

Data is based on aggregate user requests, excluding status checks.

Open in Console

Open FLUX.2 [klein] 9B KV in Console with a prompt ready to edit or send.

Help me create an image with flux-2-klein-9b-kv using image-edit at /v1/images/edits. Show the result, cost, and any limits I should know.

Use cases

Best for

Text to Image

Interactive image editors
Back a tool where users drag in several reference photos and expect an edit to come back quickly.
Batch edits on one reference set
Apply many instructions to the same references, such as outfit changes for one model photo.
Fast multi-image compositing
Place subjects from several photos into one scene without waiting on repeated reference encoding.
Prototyping edit features
Test how an edit flow feels at low latency before choosing a larger model for final renders.

Prompt examples

Combine the person in image 1 with the jacket in image 2 and the street background in image 3, keep the lighting consistent.

Change only the wall color in the reference room photo to sage green and leave the furniture untouched.

Put the sneaker from image 1 on a concrete plinth, studio lighting, grey paper background.

FAQ

What does KV mean in FLUX.2 [klein] 9B KV?

KV stands for key-value cache. The model computes attention keys and values for the reference images in the first denoising step and reuses them in the following steps instead of recomputing them each time.

How much faster is it than the standard klein 9B?

Black Forest Labs states up to 2.5 times faster for multi-reference editing. The gain depends on how many reference images you send, and text-only generation sees no such benefit.

Does it give up any quality or features?

The model card says it retains the capabilities of the standard klein 9B, including four-step text-to-image and editing. Compare outputs on your own images, since caching changes how the references are processed.

When should I choose it over the standard 9B?

Choose it when most requests carry one or more reference images and latency matters. For prompts without references, the standard klein 9B behaves the same and is the simpler pick.

Can I use it commercially?

The released weights fall under the FLUX Non-Commercial License. Commercial use of the self-hosted checkpoint needs a separate license from Black Forest Labs, so confirm terms first.

How much does FLUX.2 [klein] 9B KV cost?

On TokenLab, FLUX.2 [klein] 9B KV costs $0.00078 per request. The pricing table above shows the full breakdown. Rates depend on the billing unit, specification, and usage. Compare matching conditions in the model's detailed pricing; a single rate does not determine the total cost.

Which endpoint should FLUX.2 [klein] 9B KV use?

Use https://api.tokenlab.sh/v1/images/edits for FLUX.2 [klein] 9B KV. The request example below shows the matching code shape.

Which operations does FLUX.2 [klein] 9B KV support?

FLUX.2 [klein] 9B KV supports Image editing, Text to image. Select an operation above to see its endpoint and request example.