Head to head: Kling Image vs Flux 3 Image
Kling Image vs Flux 3 Image
By Ryan Merket · Published
We compared the two image models across eight varied prompts, from art direction and storefront scenes to precise reflections, actions, and rendered text. The results show how reliably each model turns detailed instructions into coherent images.
Flux 3 Image takes this matchup by a wide margin: it scored 67.8 to Kling Image’s 57.9, winning seven tasks outright with one tie. The statistical verdict is decisive, and the task results explain why: Flux was more dependable at satisfying several specific requirements at once without losing compositional polish. That advantage shows up across very different briefs. Flux made the requested ukiyo-e cues and dawn setting read clearly, assembled the boutique’s rain-soaked details into a legible scene, and handled the chrome teapot’s reflections more distinctly. It also showed the crucial sandal-tying action more clearly, even though its crop missed the full-body brief. In the text test, both models rendered the lines correctly, but Flux paired them with the cleaner, more restrained layout. Kling had real strengths, particularly its dramatic lighting in the knit still life and the stronger sense of relief in the missed-train portrait. That portrait was the only tie: Kling conveyed the emotion and close-up framing, while Flux supplied the more convincing garment bag and readable label. Neither was flawless, but this was a genuine trade-off rather than a Kling win hidden by the aggregate. **Final call: Flux 3 Image wins.** Its edge is not a single standout image; it is the consistency of its prompt execution across styles, object relationships, reflections, and text. Kling can deliver atmosphere, but Flux more often gets the whole brief right.
Bauhaus knit palette
A clean high-fashion still life of folded knitwear and accessories arranged on a matte display plinth: a cobalt turtleneck, mustard socks, ivory scarf, charcoal beret, and rust leather belt, rendered in a minimalist studio editorial style with hard side light and precise shadows; restrict every visible element strictly to the named Bauhaus palette of cobalt blue, mustard yellow, rust red, charcoal black, and ivory white, with no stray off-palette colors anywhere in the image, 16:9.
OpenAI: GPT-6 Luna: Model B presents all five requested items clearly in a polished, balanced editorial arrangement, with colors that closely match the named palette. Model A has appealing hard side light, but the hanging scarf and muted gray-taupe setting make the composition less cohesive and the palette less strictly controlled. (Second judge pass, order swapped — scores are the average of both: Model B presents all five requested pieces clearly in a polished, balanced studio still life, with colors closer to th | Anthropic: Claude Fable 5: Model A delivers the requested matte display plin…
Named art style
A ukiyo-e woodblock print of a fishing boat riding a large cresting wave at dawn, faithful to the flat color planes, bold outlines, and stylized foam of the tradition, muted indigo and cream palette.
OpenAI: GPT-6 Luna: Model B more fully evokes ukiyo-e through its stylized foam, bold outlines, dawn sun, and layered woodblock-like composition, while clearly showing a fishing crew aboard the boat. Model A has a striking wave and strong palette, but its dawn setting is less explicit and the scene feels less richly composed. (Second judge pass, order swapped — scores are the average of both: Model B more fully matches the prompt, with a fishing crew riding a cresting wave, a clear dawn sun, and a muted indigo-and- | Anthropic: Claude Fable 5: Model A delivers a clean, muted indigo-and-cream…
Boutique window in rain
A cinematic street-fashion scene outside an independent boutique called Virelle Noon during a light rain: a mannequin in a silver trench coat stands behind a tall curved glass window while a real model in a plum wool suit and transparent umbrella pauses on the sidewalk; the shop glass, puddles, and chrome door handle must all show physically plausible reflections and refractions of the street, clothing, signage, and passing scooter, with correct angles and transparency; ultra-realistic photography, blue-hour lighting, slightly wide composition, 16:9.
OpenAI: GPT-6 Luna: Model B presents the boutique name clearly and brings the curved display window, silver trench, plum-suited model, umbrella, rain, and passing scooter together in a polished, balanced wide composition. Model A has a stronger blue-hour mood and a prominent chrome handle, but its oversized cropped mannequin and less legible storefront make the scene feel less cohesive. (Second judge pass, order swapped — scores are the average of both: Model B more completely matches the requested boutique scene, | Anthropic: Claude Fable 5: Model B nails prompt specifics — a clearly legible…
Relief after missed train
A close-up pastel storybook illustration of a fashion intern on an empty provincial train platform, clutching a garment bag labeled "Aster Row Sample Set 7" and suddenly laughing with visible relief after realizing the departing train was the wrong one; the emotion must read instantly as relieved amusement, not generic happiness, through eyes, eyebrows, mouth, and posture; she wears a mint trench over a raspberry slip dress, wind lifting a few loose curls, warm sunrise rim light, soft painted textures, centered composition.
OpenAI: GPT-6 Luna: Model A captures the requested laughing relief, centered close-up, sunrise light, outfit, windblown curls, and legible label, though the bag reads more like a tote than a garment bag. Model B renders the garment bag and label well, but the expression reads as worry or alarm rather than relieved amusement, and the framing is less close-up. (Second judge pass, order swapped — scores are the average of both: Model B better captures the close-up brief, readable garment-bag label, and the departing t | Anthropic: Claude Fable 5: Model B delivers the actual garment bag with the…
Negation
A cozy reading nook with an armchair, a stack of books, and a mug of tea by a window — with absolutely NO plants, NO lamps, and NO artwork or picture frames anywhere in the frame. Warm afternoon light, 16:9.
OpenAI: GPT-6 Luna: Model B presents a fuller, especially inviting reading nook, with a clear stack of books and steaming tea beside the window; Model A also follows the requested setup but has fewer books and places the mug on the chair. Neither image shows indoor plants, lamps, or artwork, and both have warm afternoon light. (Second judge pass, order swapped — scores are the average of both: Model B more clearly presents the full reading nook, with the armchair, stacked books, steaming tea, and window all distinc | Anthropic: Claude Fable 5: Both Model A and Model B successfully omit plants…
Reflections & glass
A photorealistic close-up of a chrome teapot on a polished dark marble counter next to a glass of water; the window and a red apple on the counter must be correctly reflected in BOTH the chrome and the water's surface, dramatic side light, 16:9.
OpenAI: GPT-6 Luna: Model B more clearly shows the window and red apple reflected in the chrome, and the apple is also visible as a reflection through the water glass. Model A has especially dramatic lighting and polished styling, but its required reflections are less distinct. (Second judge pass, order swapped — scores are the average of both: Model B more clearly depicts a chrome teapot with the window and red apple visible in its reflections, while Model A’s teapot reads as warm-toned metal and the required refl | Anthropic: Claude Fable 5: Model B clearly fulfills the core reflection requ…
Runway tailor tying sandal
A sharp editorial fashion photograph of a young tailor-model standing full-body on a narrow backstage runway, wearing a cropped charcoal vest, pleated saffron trousers, and one finished gladiator sandal while using both hands to tie the thin leather laces on the other sandal; all ten fingers must be clearly visible and anatomically correct, with natural wrists, elbows, shoulders, and realistic body proportions, no hidden or fused digits; three-quarter view, camera at knee height, crisp studio detail, soft sidelighting from stage bulbs, faint garment racks in the background, 16:9.
OpenAI: GPT-6 Luna: Model B clearly shows the model using both hands to tie one gladiator sandal while the other is finished, with backstage racks and runway lighting; Model A has a stronger centered runway view but the tying action is unclear. Model B’s pose and cropped framing fall short of the full-body brief, and neither image clearly shows all ten fingers. (Second judge pass, order swapped — scores are the average of both: Model B clearly shows the model bending to tie the laces on one sandal while wearing the | Anthropic: Claude Fable 5: Model B correctly depicts the core action — bent…
Legible multi-line text
A minimalist event poster with three lines of crisp, correctly-spelled text stacked and centered: 'NIGHT MARKET' large on top, 'Fridays · 6–11pm' in the middle, 'Riverside Pier 4' at the bottom, on a deep navy background, clean sans-serif, subtle grain.
OpenAI: GPT-6 Luna: Model A makes the headline especially prominent, but its background is much more heavily distressed than the requested subtle grain and the time uses a hyphen. Model B has crisp, correctly rendered text, restrained texture, and a balanced minimalist layout, though it leaves generous empty space around the text. (Second judge pass, order swapped — scores are the average of both: Model B closely matches the minimalist brief, with all three lines correctly rendered, centered, and crisp against a su | Anthropic: Claude Fable 5: Both render all three lines correctly, but Model…
Matchup powered by OpenRouter.