Flux.2 Klein
A lightweight 4B parameter FLUX variant. Fast 4-step generation with low VRAM requirements.
- Developer: Black Forest Labs
- License: Apache 2.0
- HuggingFace: black-forest-labs/FLUX.2-Klein
Variants
| Model | Steps | Size | Notes |
|---|---|---|---|
flux2-klein:q8 | 4 | 4.3 GB | Good quality |
flux2-klein:q6 | 4 | 3.4 GB | Better quality |
flux2-klein:q4 | 4 | 2.6 GB | Smallest FLUX |
flux2-klein:bf16 | 4 | 7.8 GB | Full precision 4B |
Defaults
- Resolution: 1024x1024
- Guidance: 1.0
- Steps: 4
Flux.2 Klein-9B
A larger 9B parameter FLUX variant. Distilled for fast 4-step generation with higher quality than the 4B Klein. Uses a Qwen3-8B text encoder (hidden_size=4096) vs Klein-4B's Qwen3-4B (hidden_size=2560).
- Developer: Black Forest Labs
- License: Non-Commercial
- HuggingFace: black-forest-labs/FLUX.2-klein-9B (gated — requires HuggingFace license acceptance)
Variants
| Model | Steps | Size | Notes |
|---|---|---|---|
flux2-klein-9b:q8 | 4 | 10 GB | Good quality |
flux2-klein-9b:q6 | 4 | 7.9 GB | Better quality |
flux2-klein-9b:q4 | 4 | 5.9 GB | Smallest 9B |
flux2-klein-9b:bf16 | 4 | 18 GB | Full precision, gated, 2 shards |
Defaults
- Resolution: 1024x1024
- Guidance: 1.0
- Steps: 4
Note: GGUF quantized variants (Q4/Q6/Q8) use ~6-10GB VRAM. The BF16 variant requires ~18GB VRAM, is gated on HuggingFace, and requires license acceptance before download.
Recommended Dimensions
| Width | Height | Aspect Ratio |
|---|---|---|
| 1024 | 1024 | 1:1 (native) |
| 1024 | 768 | 4:3 |
| 768 | 1024 | 3:4 |
| 1024 | 576 | 16:9 |
| 576 | 1024 | 9:16 |
| 768 | 768 | 1:1 |
Using non-recommended dimensions will trigger a warning. All values must be multiples of 16.
Example
Flux.2 Klein Q8 — 4 steps, seed 100:
mold run flux2-klein:q8 \
"A minimalist zen garden with raked sand patterns, \
a single cherry blossom tree, morning mist" \
--seed 100
Flux.2 Klein BF16 — 4 steps:
mold run flux2-klein:bf16 \
"a majestic owl perched on a mossy branch in a moonlit forest"
Flux.2 Klein-9B Q4 — 4 steps, seed 999:
mold run flux2-klein-9b:q4 \
"A glass bottle ship inside a stormy ocean wave, \
dramatic lightning, hyperrealistic macro photography" \
--seed 999
Architecture
Flux.2 Klein uses a Qwen3 text encoder (BF16 or GGUF, layers 9/18/27), a shared modulation transformer (BF16 or GGUF), and a BN-VAE decoder. Klein-4B uses Qwen3-4B (hidden_size=2560), Klein-9B uses Qwen3-8B (hidden_size=4096). GGUF variants keep weights quantized in VRAM with on-the-fly dequantization per matmul, minimizing memory usage.