FLUX.1 [schnell] Black Forest Labs vs GPT Image 1 OpenAI

FLUX.1 [schnell]

• + Excellent rendering of light and reflections on the glass and wood
• + High visual quality with vibrant colors
• + Captures the 'plant partially visible through glass' instruction well

• − Added an extra blue glass sphere on top of the book that was not requested
• − The blue sphere appears to be floating inside the cube without gravity or support

GPT Image 1

• + Strict adherence to the prompt with no extra objects
• + More realistic glass texture with slight green tint on the edges
• + Clean composition and accurate lighting

• − The plant is mostly behind the cube but less visible 'through' the glass refraction compared to Image A
• − The sphere appears to be floating slightly above the bottom surface

FLUX.1 [schnell]

• + Excellent depiction of wet pavement and light rain reflections
• + High resolution with sharp details on the bicycle and man's face
• + Good use of color with the red bicycle and vest pops against the grey street

• − Fails to include the requested motion blur from passing cars
• − The bicycle geometry is slightly warped, specifically the handlebars and pedals
• − The background cars look static and parked rather than passing by

GPT Image 1

• + Successfully captures the requested crouched 'repairing' pose
• + Includes subtle motion blur in the background traffic as requested
• + Achieves a very realistic, non-stylized skin texture and candid feel

• − The framing is very tight, cutting off much of the bicycle
• − Minor anatomical issues with the fingers merging near the gears
• − Slightly lower global contrast compared to Model A

FLUX.1 [schnell]

• + extremely sharp detail on the iris and skin texture
• + very dramatic and intense facial expression
• + strong lighting contrast that highlights facial features

• − armor is cut off by the tight crop, losing the 'ornate' feel
• − braids appear somewhat thick and rope-like rather than fine
• − lacks the requested 'bokeh sparks' in the background

GPT Image 1

• + perfectly captures the ornate engraved plate armor as requested
• + excellent execution of bokeh sparks and warm torchlight atmosphere
• + most accurate depiction of battle-worn skin with realistic dirt and faint scars

• − the eyes are slightly less luminous than model a
• − overall image is a bit darker, which obscures some detail in the hair

FLUX.1 [schnell]

• + Strong minimalist layout that feels like a real physical menu.
• + Includes all three requested sections (Appetizers, Pizza, Mains).
• + Features a distinct grid arrangement for the food photos.

• − Text is largely gibberish or extremely distorted.
• − Some food photos look repetitive or low quality.
• − Includes a typo/strange word 'ORFEFUS' in a header.

GPT Image 1

• + Excellent food photography quality and realism.
• + Text is highly legible with clear prices and subheaders.
• + Successfully incorporates vibrant accents with the colored square icons.

• − Fails to include a specific section for 'Mains', instead mixing it into the 'Pizza' column.
• − Repeats the same placeholder description multiple times.
• − The grid layout is slightly uneven due to the text length.

FLUX.1 [schnell]

• + Excellent photorealistic lighting and texture on the burger bun and cheese.
• + Creative use of flying croutons and splatters to convey motion.
• + High-quality rendering of the fiery background embers.

• − Failed the text objective by clipping the 'M' in 'Magic' and duplicating the price incorrectly.
• − The burger is largely assembled rather than 'exploded' into mid-air components as requested.

GPT Image 1

• + Perfectly followed the 'exploded burger' instruction with clearly separated layers.
• + Exactly rendered the requested fiery glow effect on all text elements.
• + All prompt-required text is present, legible, and correctly placed.

• − The price in the starburst is missing the '6' before the decimal point (€.99).
• − The overall composition is somewhat static despite the exploded view.

FLUX.1 [schnell]

• + The text has a natural, imperfect handwritten feel.
• + Captures the chalkboard aesthetic well with high-contrast white on black.

• − Poor spelling throughout the menu (e.g., 'Taffle Mushmnctiomn', 'Octtoopus').
• − Fails to render the title in 'elegant cursive' as requested, using a basic print style instead.
• − The date text is truncated to 'Pril'.

GPT Image 1

• + Excellent text rendering with no spelling errors in the main items.
• + Realistic chalk texture with visible grain and slight pressure variations.
• + Follows the prompt details closely, including the specific date and menu items.

• − The title is in a stylized print rather than the requested 'elegant cursive'.
• − The prices for the cookies at the bottom are slightly cut off/incomplete.

FLUX.1 [schnell]

• + Successfully followed the difficult spatial constraint of placing the horse on top of the astronaut.
• + High cinematic lighting with a pleasing warm glow from the planet.
• + Creative and surreal interpretation that matches the prompt's intent.

• − Anatomical glitch featuring a horse with two heads/necks.
• − The astronaut's pose is somewhat nonsensical and disconnected from the backpack.

GPT Image 1

• + Very high level of detail on the space suit and horse's mane.
• + Good composition with a clear sense of atmosphere and lighting.
• + Clean anatomical rendering of both the horse and the astronaut.

• − Completely failed the negative constraint/spatial instruction for the horse to be on top.
• − Produces a generic 'astronaut riding a horse' image despite the specific prompt instruction.

FLUX.1 [schnell]

• + Excellent fur texture rendering
• + Vibrant and cinematic lighting that fits the NYC night theme
• + Clear text on the cap and interior visor

• − The paws are not both on the steering wheel correctly
• − The capybara's head shape is slightly distorted to look more like a dog-capybara mix

GPT Image 1

• + Follows the instruction for both front paws on the steering wheel better
• + Perfectly captures the bored, indifferent expression of the human passenger
• + Highly accurate capybara facial anatomy

• − Lighting is a bit flat compared to the other model
• − The texture of the paws is a bit creepy and less realistic than the face

FLUX.1 [schnell]

• + Successfully captures a vibrant, high-contrast glow on the jack-o-lantern.
• + Includes clear, illustrative elements like the twisted trees and bats as requested.

• − Several major spelling and grammatical errors in the text (e.g., 'You invited to a a night', 'firiichts', 'butistigtion').
• − Poor layout of event details with redundant and incorrect time fields (Time: 17.2026).

GPT Image 1

• + Perfect text rendering for all requested strings, including the banner and event details.
• + Superior composition that feels like a cohesive, vintage gothic poster with a gritty parchment texture.
• + Elegant used of the thorny border and cobwebs that perfectly matches the prompt.

• − Lighting is slightly more muted compared to the requested 'cinematic' glow in Model A.

FLUX.1 [schnell]

• + Excellent high-clarity 3D rendering with realistic PBR properties.
• + Perfectly centered minimal diorama aesthetic.

• − Missed the 'SUSHI' text requested in the prompt.
• − Low contrast on top-center text makes it hard to read.

GPT Image 1

• + Followed all text instructions including 'JAPAN' and 'SUSHI'.
• + Clean isometric composition with attractive clay-like textures.

• − Added extra elements like chopsticks and ginger not specifically requested.
• − The lighting is slightly flat compared to the other model.

FLUX.1 [schnell]

• + Excellent fur texture and lighting details on the grass
• + Good color saturation and vibrance

• − Failed to generate a distinct bunny, creating a weird cat/bunny hybrid instead
• − The anatomy of the two center animals is merged and confusing
• − Static posing feels less like 'chasing' and more like a portrait

GPT Image 1

• + Successfully included all four distinct species listed in the prompt
• + Captures the action of 'playfully chasing' and 'tumbling' much better
• + Beautiful rendering of god rays and morning dew consistent with the prompt

• − The fox's front paws look a bit structurally simplified
• − The butterflies are less defined than in Model A

FLUX.1 [schnell]

• + Clean vector emblem style
• + Well-balanced circular composition

• − Serious spelling errors in the main name ('FRAMILAN')
• − Incorrect date ('7720' instead of '1720')
• − Missing the requested steam element

GPT Image 1

• + Perfect text rendering for name and date
• + Includes the requested steam and banner elements
• + Excellent texture application for a vintage look

• − Uses a black background instead of the requested light/cream background
• − Slightly less 'minimalist' than Model A's linework

FLUX.1 [schnell]

• + Clean layout with a high-end vector feel
• + Good use of negative space and composition

• − Text is complete gibberish and ignores the specific step labels
• − The rocket design is generic and doesn't resemble a Saturn V
• − Fails to clearly represent the sequential 6-step infographic structure requested

GPT Image 1

• + Excellent adherence to the 6 steps with relevant icons for each
• + Text rendering is highly accurate and readable
• + Incorporated the specific mission crew and landing site names accurately

• − The layout is a bit cluttered with uneven spacing between labels
• − The Saturn V icon is slightly simplified compared to the rest of the professional vector art