Black Forest Labs
FLUX.2 Klein 4BOpen
Compact 4B FLUX.2 distilled for on-device and edge image generation.
Intelligence
86.5 #38
AAQI 87.0
Output Speed
90 t/s
ITL 11.1 ms/tok
Blended price (3:1)
$0.01 /1M
In $0.01/M · Out $0.01/M
Context
32K
≈ 80 A4 pages · max out 4,096
Quality Domain Radar
6-Axis Frontier Evaluation vs Category Average & Flagship
FLUX.2 Klein 4B
Claude Mythos Preview (#1)
Average (68)
Radar axes are approximated from category scores where direct benchmark measurements are missing — see Methodology.
Strengths & Trade-offs
Where it wins
- Runs fully on-device — no network, no fees
- ~6GB VRAM at Q4 — fits entry GPUs
- Fastest FLUX.2 tier at 90 img/s-class throughput
Trade-offs to watch
- Noticeably below Pro on fine detail and text
- Low-resolution ceiling vs. 32B tiers
- Struggles with long, complex prompts
Killer Use Cases
On-device apps
Offline image generation inside mobile/desktop apps.
Edge & IoT rendering
Local synthesis on embedded hardware.
Drafting loops
Cheap instant previews before a Pro render.
Developer Quickstart & API
Set
HF_TOKEN first — get an API keycurl https://router.huggingface.co/v1/chat/completions \
-H "Authorization: Bearer $HF_TOKEN" \
-H "content-type: application/json" \
-d '{
"model": "black-forest-labs/FLUX.2-klein-4b",
"messages": [{"role": "user", "content": "Refactor this function for O(n) time and explain the change."}]
}'import os
from openai import OpenAI
client = OpenAI(
base_url="https://router.huggingface.co/v1",
api_key=os.environ["HF_TOKEN"],
)
resp = client.chat.completions.create(
model="black-forest-labs/FLUX.2-klein-4b",
messages=[{"role": "user", "content": "Refactor this function for O(n) time and explain the change."}],
)
print(resp.choices[0].message.content)const res = await fetch("https://router.huggingface.co/v1/chat/completions", {
method: "POST",
headers: {
Authorization: `Bearer ${process.env.HF_TOKEN}`,
"content-type": "application/json",
},
body: JSON.stringify({
model: "black-forest-labs/FLUX.2-klein-4b",
messages: [{ role: "user", content: "Refactor this function for O(n) time and explain the change." }],
}),
});
console.log((await res.json()).choices[0].message.content);Official Resources & Links
API deployment
Compare hosts & latency
Throughput, TTFT and cache pricing across Groq, Cerebras, Together, DeepInfra, Fireworks, Bedrock and Azure.
GroqCerebrasTogether AIDeepInfraFireworks AIAWS BedrockAzure AI
Scores by Source
Category Scores
Benchmark Results
Head-to-head Live Matchups
Live Specs & Architecture
Under the hood
FLUX.2 Klein 4B — distilled on-device diffusion
- 4B distilled transformer — runs on phones and laptops
- Sub-second generation at 512–1024px
- Same prompt interface as Pro/Dev tiers
- Weights optimized for 6GB memory footprints
Release date
Oct 14, 2026
Architecture
FLUX.2 Klein 4B — distilled on-device diffusion
Parameters
4B
License
Open Weights
Context Window
32,000 tokens
Max Output
4,096 tokens
Modalities
text + image
Cache Pricing
read $0.004 · write $0.0125 /1M
Reasoning
—
Input Price
$0.01/M
Output Price
$0.01/M
Output Speed
90 t/s
Time to First Token
—
Data Updated
Oct 22, 2026
Hardware & Local Inference
Data: Convly AI (CC BY 4.0) ↗VRAM (4-bit Q4)
3 GB
Min GPU
Any 6GB GPU
Estimated monthly bill
Estimated monthly spend
$0.00
Calculating savings vs GPT-4o...
Cite & embed
GitHub Markdown badge
Loading badge markdown...BibTeX
@misc{llmpodium2026flux2klein4b,
title = {Benchmark Analysis and Intelligence Evaluation of FLUX.2 Klein 4B},
author = {LLMPodium Research},
year = {2026},
url = {https://llmpodium.com/models/flux-2-klein-4b}
}Embed Score Badge
Add a live Podium Score badge to your GitHub or Hugging Face README — it links back to the live ranking.
Preview
Markdown
Loading badge snippet...HTML
Loading badge snippet...More from Black Forest Labs
Alternative Models
Closest alternatives to FLUX.2 Klein 4B based on capabilities, pricing, and benchmark scores.