Black Forest Labs' open-weights image generation model with frontier performance, available for non-commercial local deployment
Settled by community votes across 3 shared challenges, with an AI judge weighing in on each.
FLUX.2 [dev]
24.5
arena score
#17 of 44 in Text-to-Image
Skill signature
Not enough comparable category data
The chart appears once both models have ratings across at least three shared arena categories.
Imagen 4.0 Ultra Generate 001
22.3
arena score
#28 of 44 in Text-to-Image
Vote tally
Where the votes landed
FLUX.2 [dev]
40.0%
win rate
Ties
0.0%
Imagen 4.0 Ultra Generate 001
60.0%
win rate
40.0%
0.0% ties
60.0%
Shared challenges
3
Challenge by challenge
The strongest take from each model on every shared challenge, with the AI judge's read.
Geometric Composition
Text-to-Image“A glass cube on a wooden table. Inside the cube is a small blue sphere. On top of the cube sits a red book. A green plant is behind the cube, partially visible through the glass. Soft window light from the left.”
FLUX.2 [dev]
Imagen 4.0 Ultra Generate 001
AI Judge Analysis
FLUX.2 [dev]
- + The glass cube looks realistic and tangible, resembling a glass brick.
- + Correctly places a blue sphere inside the cube resting on the bottom surface.
- + Excellent refraction of the plant through the glass walls.
- − The sphere appears to be slightly floating or off-center rather than firmly grounded.
- − The book texture is a bit generic and simple.
Imagen 4.0 Ultra Generate 001
- + High visual clarity with sharp edges on the glass and detailed book text.
- + The plant is clearly visible behind the cube as requested.
- + The lighting and shadows on the wooden table are well-rendered.
- − The blue sphere is physically impossible, appearing to float in the dead center of a solid glass block.
- − The reflection/refraction within the cube is confusing, showing a ghosting effect of the sphere.
Verdict: FLUX.2 [dev] followed the prompt more realistically, placing the sphere at the bottom of the glass container. While Imagen 4.0 Ultra Generate 001 has higher detail in the book and textures, it rendered the sphere floating in the middle of the glass, which looks like a digital artifact rather than a physical object inside a vessel.
Candid Street Photography
Text-to-Image“A candid street photo of an elderly Japanese man repairing a red bicycle in light rain, reflections on wet pavement, shallow depth of field, 50mm lens, natural skin texture, imperfect framing, motion blur from passing cars, cinematic but realistic, no stylization.”
FLUX.2 [dev]
Imagen 4.0 Ultra Generate 001
100% wins
0% ties
0% wins
AI Judge Analysis
FLUX.2 [dev]
- + Excellent adherence to the motion blur request for passing cars.
- + Very realistic skin texture and age-appropriate features.
- + Strong technical execution of the wet pavement reflections and light rain visualization.
- − The bicycle structure is physically incoherent with tangled wires and impossible frame geometry.
- − Subject is positioned dangerously close to high-speed blurred traffic, creating a logical inconsistency.
Imagen 4.0 Ultra Generate 001
- + Highly detailed skin texture and facial features with a very realistic photographic look.
- + The bicycle components, while simplified, are much more coherent and recognizable than Model A.
- + Excellent rendering of water droplets on the man's jacket and the ground.
- − Failed to include the requested motion blur of passing cars, as the background vehicles are static.
- − The white spots on the ground look more like petals or litter than rain reflections/splashes.
Verdict: FLUX.2 [dev] followed the complex motion blur and street-photography framing prompts much better, though the bicycle itself is a mess of AI artifacts. Imagen 4.0 Ultra produced a much cleaner and more believable subject and prop, but completely ignored the 'motion blur from passing cars' instruction. FLUX.2 [dev] is preferred for capturing the requested atmosphere and technical motion effects, despite the structural failures of the bike.
Adorable Baby Animals in Sunny Meadow
Text-to-Image“Hyper-photorealistic scene of fluffy baby animals—a golden retriever puppy, tabby kitten, baby bunny, and red fox kit—with big expressive eyes and ultra-detailed soft fur, playfully chasing butterflies and tumbling together in a lush wildflower meadow, warm golden sunrise light with god rays and dew sparkles, joyful wholesome vibe, 8K masterpiece.”
FLUX.2 [dev]
Imagen 4.0 Ultra Generate 001
25% wins
0% ties
75% wins
AI Judge Analysis
FLUX.2 [dev]
- + Excellent hyper-photorealistic textures, particularly the backlighting on the fur.
- + Very natural composition with realistic lighting and 'god rays' that feel integrated into the scene.
- + Accurate species representation with high anatomical fidelity for all four animals.
- − The animals are sitting relatively still rather than 'chasing' or 'tumbling' as requested.
- − The 'dew sparkles' are present but subtle compared to the other model.
Imagen 4.0 Ultra Generate 001
- + Strong adherence to the action keywords, showing the animals playfully 'chasing' and standing on hind legs.
- + Highly vibrant and whimsical colors that enhance the 'joyful' vibe.
- + Clear inclusion of all prompt elements, including prominent dew drops and many butterflies.
- − Lacks photorealism, leaning more toward a digital illustration or 3D render style.
- − The kitten's anatomy and pose look slightly awkward and less realistic than the other animals.
- − Over-sharpened details make the fur look somewhat artificial.
Verdict: FLUX.2 [dev] produces a significantly more realistic and aesthetically pleasing 'masterpiece' with sophisticated lighting and texture, though the animals are more stationary than the prompt suggested. Imagen 4.0 Ultra captures the playful action and 'chasing' aspect much better, but the visual style feels more like a cartoon or stock illustration rather than the requested hyper-photorealistic scene. FLUX.2 [dev] is the winner for its superior visual quality and expert handling of light and fur.
Next steps
Explore each model
Google's Imagen 4.0 Ultra model offering the highest fidelity and resolution for professional-grade image generation