Black Forest Labs' open-weights multimodal flow transformer for in-context image generation and editing, available for non-commercial use with character consistency and style transfer capabilities
Settled by community votes across 4 shared challenges, with an AI judge weighing in on each.
FLUX.1 Kontext [dev]
16.5
arena score
#58 of 62 in Text-to-Image
Skill signature
· Text-to-Image
Qwen Image Edit 2509
7.8
arena score
#32 of 32 in Image Editing
Vote tally
Where the votes landed
FLUX.1 Kontext [dev]
0%
win rate
Ties
0%
Qwen Image Edit 2509
0%
win rate
Shared challenges
4
Challenge by challenge
The strongest take from each model on every shared challenge, with the AI judge's read.
Bald man challenge
Image EditingEdit instruction
“Give the person a full, thick head of natural hair with realistic texture, density, and a natural hairline. Preserve facial features and lighting.”
FLUX.1 Kontext [dev]
Qwen Image Edit 2509
AI Judge Analysis
FLUX.1 Kontext [dev]
- + Successfully added thick hair with a natural texture
- + Maintains strong facial detail and sharpness
- − Significantly altered the facial features, making the man look younger and thinner
- − Changed the style and color of the eyeglasses
- − Overall identity preservation is poor
Qwen Image Edit 2509
- + Excellent preservation of the original facial features and identity
- + Maintains the original glasses, lighting, and skin texture
- + Added hair fits the volume and density requested in the prompt
- − The hair texture has some digital-looking waves that appear slightly artificial at the top
- − The hair color is slightly cooler/darker than the beard, creating a minor mismatch
Verdict: While FLUX.1 Kontext [dev] produced a very clean image, it failed at the core task of identity preservation, effectively generating a new person who looks different from the source. Qwen Image Edit 2509 successfully added the hair while keeping the exact face, glasses, and background of the original image intact, making it a much more successful edit.
Over-the-top cartoon caricature
EditingEdit instruction
“Create a caricature of me and my job. Make it exaggerated and humorous, incorporating my profession as a tv show anchor and my love for dogs and hockey.”
Source
FLUX.1 Kontext [dev]
Qwen Image Edit 2509
AI Judge Analysis
FLUX.1 Kontext [dev]
- + Maintains a strong facial resemblance to the subject in a comic style.
- + Includes a TV screen and a dog as requested.
- + Preserves the specific denim shirt and pose from the source image accurately.
- − Completely fails to include any hockey elements.
- − The background elements like the floating pictures are generic and lack detail.
- − The text 'JOB' is a very literal and uncreative interpretation.
Qwen Image Edit 2509
- + Successfully incorporates all elements: TV anchor desk, dogs, and hockey players.
- + Excellent caricature style with exaggerated features that still look like the subject.
- + Highly creative composition that merges the news set with a hockey rink.
- − The text in the speech bubble and on the screen is gibberish.
- − The dog is very small and lacks detail compared to the rest of the image.
Verdict: Qwen Image Edit 2509 is the clear winner as it successfully incorporated all three requested thematic elements (TV anchor, dogs, and hockey) into a cohesive and creative caricature. FLUX.1 Kontext [dev] completely missed the hockey requirement and provided a much simpler, less imaginative scene, despite better text rendering of the word 'JOB'.
Studio Ghibli Anime Style
EditingEdit instruction
“Transform this photo into a Studio Ghibli–inspired illustration. Use soft pastel colors, hand-painted textures, gentle lighting, dreamy backgrounds, and a warm, nostalgic mood”
Source
FLUX.1 Kontext [dev]
Qwen Image Edit 2509
AI Judge Analysis
FLUX.1 Kontext [dev]
- + Excellent preservation of the original subjects' poses and clothing details
- + Clean, high-resolution line work and vibrant colors
- + Strong character facial expression fidelity to the original meme
- − Lean more toward generic modern anime style than Studio Ghibli's specific aesthetic
- − Colors are a bit too saturated for the requested 'soft pastel' palette
Qwen Image Edit 2509
- + Perfectly captures the Studio Ghibli soft pastel color palette and warm lighting
- + Features hand-painted paper textures that align with the requested style
- + Successfully translates the facial features into a more Ghibli-esque character design
- − The halftone-like dot texture in some areas clashes slightly with the hand-painted request
- − Slightly less crisp than Model A
Verdict: Both models did an exceptional job of preserving the composition and characters of the original 'Distracted Boyfriend' meme while applying an artistic filter. FLUX.1 Kontext [dev] produced a cleaner, more vibrant anime illustration, but Qwen Image Edit 2509 followed the specific stylistic prompts (soft pastels, dreamy background, warm mood) much more accurately to the Studio Ghibli brand.
Golden Hour Stroll
Image EditingEdit instruction
“Add dynamic motion to this photo: make hair blow in the wind, add leaves flying, energetic and lively feel.”
FLUX.1 Kontext [dev]
Qwen Image Edit 2509
AI Judge Analysis
FLUX.1 Kontext [dev]
- + Excellent preservation of the original subjects' faces and textures.
- + Realistic hair movement and subtle addition of wind-blown leaves.
- + High overall image quality and coherence with the original lighting.
- − The amount of flying leaves is very minimal, almost unnoticeable.
Qwen Image Edit 2509
- + Strong adherence to the flying leaves prompt with vibrant, varied colors.
- + Noticeable and dynamic hair movement.
- − Significant distortion of the woman's face and eyes compared to the source.
- − Hair texture becomes somewhat plastic-like and less natural than the source.
Verdict: FLUX.1 Kontext [dev] is the superior model for this task because it maintains the integrity of the source image while applying subtle, realistic hair movement and adding leaves, even if the leaf count is low. Qwen Image Edit 2509 successfully adds more 'dynamic' elements like many flying leaves, but it fails the primary goal of image editing by significantly altering the woman's facial features and lowering overall realism.
Next steps
Explore each model
Alibaba's Qwen image editing model for instruction-based image modifications and transformations