FLUX.2 klein base 9B

9B base rectified flow model (paired with ~8B Qwen text encoder), BF16 weights take ~22 GB and FP8/NVFP4 quantizations require around 10 to 14 GB VRAM

• BARE MINIMUM

  NVIDIA GeForce RTX 3060 12GB

  12 GB VRAM

  ~$220 USED

  GGUF Q4 / NVFP4 quant with offloading, 20-30 steps, ~55 s/image

  
• SWEET SPOT

  NVIDIA GeForce RTX 4080 16GB

  16 GB VRAM

  ~$850 USED

  FP8 quant with partial encoder offload, 20-30 steps, ~14 s/image

  
• FULL POWER

  NVIDIA GeForce RTX 4090 24GB

  24 GB VRAM

  ~$1600 USED

  Native FP8 / BF16 pipeline fitting entirely in VRAM, 20-30 steps, ~6 s/image

  

HARDWARE PICKS & PRICES ARE RESEARCHED FROM THE LIVE WEB AND REFRESHED AUTOMATICALLY. TREAT THEM AS BALLPARK, NOT GOSPEL.

STARS ARE THE COMMUNITY SCORE FOR THAT KIND OF WORK, DOCKED FOR HOW SQUASHED THE WEIGHTS GET AT EACH TIER. TIMES ARE COMPUTED FROM RESEARCHED SPEEDS. BALLPARK, NOT GOSPEL.

Not enough discussion about FLUX.2 klein base 9B yet to call it. We found 310 posts but people did not say much either way.

249 POSTS · 61 COMMENTS · REDDIT · GITHUB · X