Head to head: Kling Image vs OpenAI: GPT-5.4 Image 2
Kling Image vs OpenAI: GPT-5.4 Image 2
By Ryan Merket · Published
This eight-task image-model matchup tests exact text, detailed scene instructions, anatomy, and visual style. The central trade-off is prompt-specific detail versus reliable widescreen framing.
OpenAI: GPT-5.4 Image 2 leads the saved results with an aggregate score of **67.3**, against **55.6** for Kling Image. It won six tasks to Kling’s two, with no ties—a clear editorial result on this sample, not a statistical claim. OpenAI’s edge shows up in several detail-heavy prompts: it rendered the requested multi-line text more faithfully, made the ukiyo-e reference more unmistakable, depicted four separate moth wings rather than whole moths, and delivered the specified cryostat reflection details. Those are meaningful wins for prompts where getting the particular objects and wording right matters more than simply making an attractive image. Kling’s two saved wins were **Negation** and **Hands & anatomy**. Its stronger adherence to widescreen framing is a recurring point in its favor, while OpenAI’s square crops were a notable mismatch in several tasks—including the cleanroom and cryostat scenes. That format weakness keeps the result from being a blanket endorsement of OpenAI’s compositions. **Verdict: OpenAI: GPT-5.4 Image 2 wins this saved task sample, 67.3 to 55.6.** It was more dependable across the range of specific visual requirements, though Kling’s framing advantage is worth noting. This assessment applies only to the saved task sample.
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 B renders all three lines correctly, including the requested time range, in a polished centered layout with subtle grain. Model A is attractive and legible, but changes the time-range punctuation and is less faithful to the requested text. (Second judge pass, order swapped — scores are the average of both: Model B renders all three requested lines accurately, including the centered dot and 6–11pm dash, on a textured deep navy background. Model A has a strong centered layout, but changes th | Anthropic: Claude Fable 5.1: Model B renders all three lines exactly as spec…
Sterile Dock Without Gloves
A blueprint-style technical drawing with subtle cyanline shading, 16:9, depicting a compact biochemistry loading dock at midnight: a stainless transfer hatch, labeled canisters of lysate buffer, a rolling rack of sealed sample trays, wall-mounted barcode scanner, floor drain, emergency shower, and a digital clock reading 00:17. The room must be completely empty of people and must explicitly contain NO gloves anywhere—no latex gloves, no nitrile gloves, no glove boxes, no discarded gloves, no glove icons on signs, nothing glove-shaped at all. Clean orthographic-meets-perspective composition, cool fluorescent lighting, tidy labeled equipment, and a sterile industrial calm.
OpenAI: GPT-6 Luna: Model A better matches the requested 16:9 blueprint-style scene and has a clean, cohesive composition, with the clock, hatch, canisters, tray rack, drain, and no gloves or people clearly present. Model B labels more of the requested equipment explicitly, especially the emergency shower, but its square infographic layout departs substantially from the requested aspect ratio and feels busier. (Second judge pass, order swapped — scores are the average of both: Model B clearly includes nearly every | Anthropic: Claude Fable 5.1: Model B delivers a genuine blueprint sheet with…
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 better matches the requested 16:9 framing while clearly showing the armchair, books, tea, and warm afternoon light; Model B is especially cozy and well composed but is square rather than widescreen. Both avoid indoor plants, lamps, and artwork. (Second judge pass, order swapped — scores are the average of both: Model B is warm and polished, but its square framing misses the requested 16:9 format. Model A better matches the aspect ratio while clearly showing the cozy chair, books, steamin | Anthropic: Claude Fable 5.1: Both images correctly omit plants, lamps, and a…
Orbital Cleanroom Perspective
A precise wide-angle 16:9 scene in a near-future satellite assembly cleanroom, rendered as ultra-clean photoreal concept art: the camera is low to the epoxy floor, looking down a long aisle of ceiling trusses toward a single vanishing point. In the foreground, a tiny hexagonal calibration cube marked Pavo-3 sits near the lens; midground, two technicians in white bunny suits wheel a waist-high nitrogen cart; background, a towering cobalt communications satellite with unfolded dish segments rises three stories under gantry lights. Relative scale must read correctly from front to back, with parallel floor seams, cable trays, and scaffold rails converging consistently, no warped walls, and every object sized believably against the humans. Bright neutral cleanroom lighting, faint reflections on the polished floor, immaculate geometric composition.
OpenAI: GPT-6 Luna: Model A more closely matches the requested 16:9 low-angle, symmetrical cleanroom aisle, with strong converging lines, polished-floor reflections, and a clearly marked foreground cube. Model B depicts the satellite’s dish segments and technicians with a cart more literally, but its square framing and less pronounced vanishing-point composition fit the prompt less well. (Second judge pass, order swapped — scores are the average of both: Model B more faithfully depicts the cleanroom aisle, towering | Anthropic: Claude Fable 5.1: Model B nails nearly every content beat—low cam…
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 convincingly evokes a traditional ukiyo-e print, with a dramatic cresting wave, fishing crew, and clear dawn setting; Model A has the closer muted indigo-and-cream palette but a less distinctive composition and weaker dawn cues. Model B’s added lettering is decorative, though not requested. (Second judge pass, order swapped — scores are the average of both: Model B more clearly depicts a fishing boat and crew riding a cresting wave, with a visible dawn sun, muted indigo-and-cream co | Anthropic: Claude Fable 5.1: Model B delivers a convincing ukiyo-e print wit…
Moth Wing Archive Detail
An extreme close-up scientific illustration in hyper-detailed natural-history gouache and ink, showing a laboratory specimen tray from the fictional Aster Vale Entomology Archive: four pinned noctuid moth wings, a strip of indigo velvet, brass calipers, handwritten tag AV-2041, and a cracked bakelite loupe resting on graph paper. The image should emphasize fine texture and detail: powdery wing scales in distinct bands of rust, ash, and celadon, tiny frayed edges, the velvet weave, oxidized scratches on brass, paper fibers, and hairline crazing in the loupe handle, all rendered crisp and unsmudged. Raking side light from the left creates microscopic shadows and reveals surface relief, with a carefully composed top-down arrangement.
OpenAI: GPT-6 Luna: Model B more closely follows the specimen-focused prompt with four separate, banded wings arranged on graph paper and convincing archive details, while Model A depicts four whole moths instead. Model A has a clear tag and dramatic cracked loupe, but Model B’s top-down composition and fine surface textures are more coherent overall. (Second judge pass, order swapped — scores are the average of both: Model B closely matches the requested top-down archive study, with four separate pinned wings, the | Anthropic: Claude Fable 5.1: Model B correctly depicts four individual pinne…
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 both hands actively tying a friendship bracelet with a plausible grip on the threads, while Model A has a warmer, more cinematic setting and matches the requested 16:9 framing. Model B’s square crop is a notable mismatch, but its hand-and-thread interaction is more legible. (Second judge pass, order swapped — scores are the average of both: Model B clearly shows the bracelet being tied, but its square crop misses the requested widescreen format. Model A has a natural, | Anthropic: Claude Fable 5.1: Model B delivers a far more convincing candid kn…
Cryostat Reflection Bench
A photorealistic 16:9 editorial science image of a polarimetry lab at dawn: on a brushed-steel bench sits a glass Dewar cryostat labeled VX-17 beside a shallow black dish of mercury, a chrome micrometer stage, and three amber reagent vials; behind them, a tall safety mirror and a rain-streaked window reveal the room from different angles. The reflections and transparency must be physically coherent: the mercury dish mirrors the underside of the cryostat and the orange vials, the chrome stage reflects nearby objects with curved distortion, the mirror shows the back of a technician in a teal lab coat writing "zirconate run 4" on a clipboard, and the window subtly refracts the exterior radio mast and interior shelf through droplets. Cool blue ambient light from the stormy window mixes with a narrow warm task lamp, producing crisp highlights, soft shadows, and believable refraction through glass walls and liquid menisci.
OpenAI: GPT-6 Luna: Model B more convincingly realizes the requested reflection setup, including the back-facing technician writing the specified phrase, the reflective mercury dish, and the labeled cryostat; Model A has the requested wide framing but misses or distorts several of these details. Model B’s main shortcoming is its square format rather than the requested 16:9, while both images have polished lighting and strong visual detail. (Second judge pass, order swapped — scores are the average of both: Model B | Anthropic: Claude Fable 5.1: Model B nails the key specifics—legible 'VX-17'…
Matchup powered by OpenRouter.