Head to head: Wan vs Flux 3 Image
Wan vs Flux 3 Image
By Ryan Merket · Published
A task-by-task comparison of two image models across demanding scenes involving precise layouts, reflective materials, anatomy, and detailed lab setups. The tests reward not just visual polish, but whether each image follows the brief.
Flux 3 Image wins the matchup on both score and consistency: **65.3 to Wan’s 57.7**, with a **limited confidence** verdict. It took six tasks, tied two, and lost none—a decisive result backed by repeated success with specific prompt constraints, not just attractive rendering. Its clearest advantages came when the brief demanded exact placement or countable details. Flux put the bedroom desk beneath the window, arranged the seven isotope vials in a loose zigzag, and kept forbidden lamps, artwork, and plants out of the reading nook. In the assay-room and cryostat scenes, it also supplied more faithful equipment and more convincing environmental reflections. The two ties show where the gap narrows. Both models produced persuasive hands-and-bracelet scenes, and neither fully nailed the requested reflections on the water in the glass-and-teapot comparison. Wan could deliver strong composition and polish—the judges noted its centered cryostat framing and compelling chemist—but Flux more often matched the requested action, objects, and spatial relationships. **Final call: Flux 3 Image is the stronger prompt-following image model in this head-to-head, with a decisive edge across the tested tasks.**
Botanist with seed trays
A realistic editorial-style portrait of a field botanist in a greenhouse genetics lab, shown full upper body at a stainless workbench, carefully transplanting rare spiral seedpods into labeled culture trays; both hands are fully visible and anatomically correct, with natural finger positions as one hand steadies tweezers and the other cups a small ceramic dish, no obscured or fused fingers, believable wrists and forearms, accurate shoulder width and posture; she wears a rolled-sleeve moss-green lab coat, a faded indigo headscarf, and clear safety glasses; soft overcast daylight through roof panels, shallow depth of field, 16:9.
OpenAI: GPT-6 Luna: Model A more clearly shows the botanist transplanting spiral seedpods into culture trays, with both hands visible and the requested tools and dish in use. Model B is similarly convincing and has cleaner labeling, but the seedpods remain in the dish while the trays are mostly empty. (Second judge pass, order swapped — scores are the average of both: Model B more convincingly delivers a realistic editorial greenhouse portrait, with clear visible hands, safety glasses, the requested clothing, and l | Anthropic: Claude Fable 5.1: Model A delivers the requested editorial look w…
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 A has a more convincing close-up teapot composition and crisp window reflections across the chrome and polished counter. Model B makes the apple and glass especially clear, but neither image convincingly shows both the window and apple reflected on the water’s surface. (Second judge pass, order swapped — scores are the average of both: Model B better matches the dramatic, dark-marble close-up and shows strong window and apple reflections across the chrome and glass. Model A is polished and | Anthropic: Claude Fable 5.1: Model B's chrome convincingly reflects the wind…
Spatial layout
A clean isometric illustration of a bedroom: a bed against the LEFT wall, a round rug centered on the floor, a desk under the WINDOW on the back wall, and a floor lamp in the FRONT-RIGHT corner. Flat-vector style, consistent perspective.
OpenAI: GPT-6 Luna: Model B more clearly places the desk directly beneath the back-wall window and the floor lamp in the front-right corner, while keeping the bed along the left wall and rug near the center. Model A has an attractive isometric layout, but its desk is offset from the window and its room perspective is less consistent. (Second judge pass, order swapped — scores are the average of both: Model B places the desk directly beneath the back-wall window and includes the requested bed, centered rug, and fron | Anthropic: Claude Fable 5.1: Model A places the bed on the left wall, center…
Cryostat reflection chamber
A photorealistic 16:9 interior of a compact polar research laboratory at blue hour, centered on a cylindrical cryostat behind a curved glass safety shield with faint frost at the edges; the shield must clearly refract the cryostat’s copper coils and reflected control panels, while a polished chrome nitrogen canister and a shallow puddle on the epoxy floor both mirror the room from their own correct angles; outside a narrow window, a wind-bent orange mast light is visible and should appear distorted through the glass layers; composition is three-quarter view from waist height, crisp realistic detail, cool cyan practical lighting mixed with one warm inspection lamp.
OpenAI: GPT-6 Luna: Model B better captures the wind-bent mast, convincing reflections in the floor puddle and chrome canister, and a detailed cryostat behind a frosted curved shield. Model A has a stronger centered composition, but its mast is upright and its floor puddle reflects less clearly. (Second judge pass, order swapped — scores are the average of both: Model B better captures the requested three-quarter laboratory view, with a clearly curved frosted shield, richly detailed equipment, convincing floor and | Anthropic: Claude Fable 5.1: Model B delivers a far more photorealistic lab w…
Hands & anatomy
A natural candid photo of a person tying a friendship bracelet, both hands clearly visible with anatomically correct fingers and a plausible grip on the threads, soft daylight, shallow depth of field, 16:9.
OpenAI: GPT-6 Luna: Model B more clearly shows a person actively tying a friendship bracelet, with both hands in a natural, plausible grip and a candid daylight setting. Model A has a strong close-up of the hands, but the extra pointed tool and tighter framing make the bracelet-making action less clear. (Second judge pass, order swapped — scores are the average of both: Model B provides a convincing candid craft scene with both hands and bracelet supplies clearly visible, while Model A’s close framing makes the han | Anthropic: Claude Fable 5.1: Model A delivers exactly what the prompt asks—t…
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 A includes prominent leafy greenery outside the window and uses a wider-than-requested aspect ratio, while Model B more closely matches 16:9 and keeps the scene free of visible plants, lamps, and artwork. Both create a warm, attractive reading nook with the requested chair, books, tea, and window. (Second judge pass, order swapped — scores are the average of both: Model B creates a warm, inviting reading nook with all the requested furnishings and no lamps or artwork; Model A does as well, | Anthropic: Claude Fable 5.1: Both outputs avoid lamps, artwork, and indoor p…
Seven isotope vials
An ultra-detailed photorealistic top-down product shot of a laboratory inventory tray on a matte black counter, containing EXACTLY seven distinct borosilicate isotope sample vials, all fully visible and individually countable, arranged in a loose zigzag with small gaps between them: one with amber liquid, one with cobalt-blue crystals, one frosted empty vial, one with a white powder pellet, one with pale green gel, one with three tiny copper beads, and one with a rolled paper label inside; no extra vials, no cropped items, no duplicates, no background clutter beyond a silver marker and a barcode card at the edge; clean neutral studio lighting, sharp focus, 16:9.
OpenAI: GPT-6 Luna: Model B better matches the requested loose zigzag and shows the white material as a distinct pellet, while keeping all seven vials and the requested edge props visible. Model A has a polished black-tray look, but its vials form a straight row and the label text is less legible. (Second judge pass, order swapped — scores are the average of both: Model B shows exactly seven distinct, fully visible vials with all requested contents, and its staggered placement comes closer to the requested loose zi | Anthropic: Claude Fable 5.1: Model B delivers exactly seven countable vials…
Noir materials test in assay room
Film-noir black-and-white photorealistic scene, 16:9, in a late-night metallurgical assay room: a tired chemist sits beside a scarred oak bench under a single tungsten desk lamp, with a brushed steel spectrometer, a cracked leather notebook, a glazed ceramic mug, folded wool coat on a chair, and a glass desiccator catching a sharp highlight; the image should feel physically believable, with convincing falloff from the lamp, deep but readable shadows, realistic specular reflections on metal, soft sheen on skin, matte grain in the wood, and distinct material responses for fabric, glass, and ceramic; low camera angle, cinematic contrast, subtle cigarette-smoke haze without stylization.
OpenAI: GPT-6 Luna: Model B more clearly conveys a working metallurgical assay room, with a spectrometer-like instrument, desiccator, coat, notebook, and convincing lamp falloff arranged in a strong cinematic composition. Model A has especially evocative smoke and a compelling chemist, but its prominent microscope is a less faithful match for the specified spectrometer. (Second judge pass, order swapped — scores are the average of both: Model B more closely matches the assay-room brief, especially with its spectrom | Anthropic: Claude Fable 5.1: Model B reads as a genuine assay lab with a con…
Matchup powered by OpenRouter.