Distilled version of Black Forest Labs' FLUX.2 [dev] outperforming it at a cheaper price. Developed by fal.ai.
Settled by community votes across 2 shared challenges, with an AI judge weighing in on each.
FLUX.2 [dev] Turbo
#4 of 44 in Text-to-Image
Not enough comparable category data
The chart appears once both models have ratings across at least three shared arena categories.
FLUX.2 [klein] 9B
#7 of 23 in Image Editing
Where the votes landed
FLUX.2 [dev] Turbo
0.0%
win rate
Ties
0.0%
FLUX.2 [klein] 9B
100.0%
win rate
Challenge by challenge
The strongest take from each model on every shared challenge, with the AI judge's read.
Chalkboard Menu
Text-to-Image“Handwritten-style chalkboard menu in a cozy café, all text rendered in the exact same realistic chalk handwriting style with natural variations in letter size, slight slant, and chalk texture — no printed or digital fonts anywhere on the board. Title at the top in elegant cursive chalk handwriting: ‘TODAY’S SPECIALS – APRIL 30, 2026’. Below it, three menu items also in the same handwritten chalk style: ‘Truffle Mushroom Risotto – $24’, ‘Grilled Octopus with Lemon & Herbs – $28’, ‘Brown Butter Chocolate Chip Cookies – $9’. At the very bottom, smaller text in the identical handwritten chalk style (slightly smaller but still clearly legible with the same handwriting characteristics): ‘All items made fresh daily • Ask about our gluten-free options’. Warm ambient café lighting, visible chalk dust and smudges, realistic handwriting imperfections, no clean printed text anywhere.”
AI Judge Analysis
FLUX.2 [dev] Turbo
- + Excellent spelling accuracy for a complex prompt and handwriting style.
- + Realistic chalk texture with convincing dust smudges and varying stroke weight.
- + Very natural-looking café environment in the background.
- − The pricing on the third item is a bit repetitive ('- 9' then '$9').
FLUX.2 [klein] 9B
- + Strong chalk texture and realistic wood grain on the frame.
- + Clear, legible handwriting that matches the requested style.
- − Significant spelling errors including 'Riott Risoto', 'Octoopus', and 'fress'.
- − Included a random repeat of the price '-$28' in the middle of a line.
Verdict: FLUX.2 [dev] Turbo significantly outperformed FLUX.2 [klein] 9B by maintaining near-perfect spelling across all menu items. While both models captured the 'handwritten chalk' aesthetic well, FLUX.2 [klein] 9B suffered from numerous typographical errors and layout hallucinations, whereas FLUX.2 [dev] Turbo produced a professional, usable menu image.
Pose & Character Mashup
Editing“Use Image 1 as the exact pose reference and Image 2 as the character reference. Recreate the person/character from Image 2 in the exact dynamic pose and body position from Image 1. Keep the exact face, hair, clothing style/details, and expression from Image 2. Match the lighting and environment of Image 1. The final image must show the character from Image 2 performing the precise action/pose from Image 1 with perfect anatomy and natural integration.”
AI Judge Analysis
FLUX.2 [dev] Turbo
- + Successfully incorporates the character's face, sunglasses, scarf, and shirt details.
- + Matches the studio lighting and yellow background of the source image.
- − Horrific anatomical failure with a second female head growing out of the character's torso.
- − Fails to replicate the specific crossed-leg pose from Image 1, resulting in a floating figure.
FLUX.2 [klein] 9B
- + Perfectly replicates the complex body pose from Image 1.
- + Integrates the character's clothing and accessories seamlessly into the action.
- + Preserves the anatomy and spatial relationship with the red stool.
- − The head angle is slightly less dynamic than the original source face but remains structurally sound.
Verdict: FLUX.2 [klein] 9B followed the instructions perfectly, successfully mapping the character from Image 2 onto the complex pose in Image 1 with high anatomical fidelity. In contrast, FLUX.2 [dev] Turbo suffered a major hallucination, including a severed/additional female head and failing to replicate the specific physical interaction with the stool.
Explore each model
Black Forest Labs' distilled 9 billion parameter image generation model with sub-second inference and multi-reference support