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
A prompt-following stress test pits Kling Image against OpenAI: GPT-5.4 Image 2 across everything from exact object counts to illustrated scenes and hand gestures.
OpenAI: GPT-5.4 Image 2 wins this matchup decisively: **64.5 to 50.4**, with a **limited confidence** verdict. It took seven of eight tasks; Kling Image’s only win was the hands-and-anatomy prompt, where its 16:9 framing and candid composition edged out OpenAI’s clearer bracelet-tying action. OpenAI’s advantage showed most in prompts where details had to be both present and legible. It delivered exactly seven countable cups, included the specified moonflowers, sage, fox, and branding in the sleep-spray poster, and made the requested room layout and aisle product sizing easier to read. Those are practical wins in instruction-following, not just polish. It wasn’t flawless. Square framing repeatedly ignored 16:9 instructions, including in the campsite, countertop, aisle, and rainy vending-machine scenes; visible gradients also undercut the restricted-palette brief. Kling more often honored the requested widescreen composition, but its wins on framing didn’t outweigh OpenAI’s stronger execution of scene contents and spatial relationships. **Final call: OpenAI: GPT-5.4 Image 2 is the stronger image model in this head-to-head—more reliable on specific visual requirements, with format compliance as its clearest recurring weakness.**
Restricted palette
A flat-vector illustration of a mountain campsite at dusk using ONLY four colors — cream, burnt orange, teal, and charcoal — with no gradients and no stray off-palette colors, balanced composition, 16:9.
OpenAI: GPT-6 Luna: Model A better matches the requested 16:9 flat-vector campsite scene, with a clear, balanced arrangement, though its turquoise flame strays from the four-color palette. Model B is polished and uses a cohesive palette, but its square format and visible background gradient conflict with the prompt. (Second judge pass, order swapped — scores are the average of both: Model B more clearly depicts a dusk campsite with the requested four-color scheme, though its square format and visible gradients viol | Anthropic: Claude Fable 5.1: Model B holds the four-color restriction almost…
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 the bracelet with a plausible grip, while Model A’s thread handling feels less clearly connected to the task. Model A better matches the requested 16:9 framing, but Model B’s close composition and soft, candid look are strong. (Second judge pass, order swapped — scores are the average of both: Model B shows the knotting action and hand grips especially clearly, but Model A better matches the requested 16:9 candid composition and presents both | Anthropic: Claude Fable 5.1: Model A delivers the requested 16:9 frame with s…
Countertop layout logic
A photorealistic product-shot scene on a pale terrazzo kitchen counter, 16:9: a matte coral electric kettle sits exactly in the center; a tiny chrome teaspoon lies directly in front of the kettle; a cobalt-blue ceramic mug is to the left of the kettle with its handle pointing left; a clear glass honey jar is to the right of the kettle; a striped tea tin is behind the kettle; a sliced lime on a small wooden coaster is between the mug and the kettle; and a folded cream dish towel is under the honey jar only, not touching any other object, shot from a slightly elevated three-quarter angle with crisp morning window light.
OpenAI: GPT-6 Luna: Model A follows the requested 16:9 layout especially well, with the tea tin behind the centered kettle and the other objects in the specified positions. Model B is polished, but its square framing and tea tin placed beside rather than behind the kettle make it less faithful overall. (Second judge pass, order swapped — scores are the average of both: Model A better matches the requested 16:9 framing and keeps the countertop arrangement clear, while Model B has a more convincingly matte kettle and | Anthropic: Claude Fable 5.1: Model B nails nearly every spatial instruction…
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 bed along the left wall, the desk beneath the back-wall window, the centered round rug, and the lamp in the front-right corner. Model B also has a more polished, cohesive flat-vector look, while Model A is starkly monochrome and more rendered than illustrated. (Second judge pass, order swapped — scores are the average of both: Model B clearly places the bed along the left wall, centers the round rug, puts the desk beneath the back-wall window, and places the lamp | Anthropic: Claude Fable 5.1: Model B delivers a true flat-vector isometric be…
Aisle perspective sizing
A hyper-detailed wide-angle retail interior, 16:9, showing a long pharmacy aisle in precise one-point perspective: in the foreground at bottom right stands a 28 cm tall amber pump bottle of hand soap labeled "Northloop Citrus," beside a 9 cm travel-size tube of mint toothpaste, while farther down the aisle a stack of boxed facial tissues and a row of laundry detergent jugs shrink consistently toward a single vanishing point at the center rear; overhead shelf lines, floor tiles, and price tags all converge cleanly, with no warped products or inconsistent scale, cool neutral store lighting, camera at standing eye level.
OpenAI: GPT-6 Luna: Model A has a striking wide-angle aisle, but the toothpaste is nearly as large as the soap bottle and the requested product categories are less clearly distinguished. Model B better conveys the specified bottle-to-tube size difference and visibly includes tissues and detergent, though its square framing misses the requested 16:9 format. (Second judge pass, order swapped — scores are the average of both: Model B renders the requested product labels and relative sizing more clearly, with a convinc | Anthropic: Claude Fable 5.1: Model A delivers the requested 16:9 wide-angle…
Neon vending rain dusk
A cinematic street photograph of a red vending machine selling canned sparkling water outside a closed neighborhood laundromat, 16:9, at neon-blue rainy night: wet pavement reflecting magenta signage, fine mist in the air, droplets beading on the machine’s plastic window, and a single umbrella leaning against the side; the scene should feel moody and urban yet intimate, lit primarily by the vending machine glow and a flickering cyan laundromat sign, with deep shadows and realistic specular reflections.
OpenAI: GPT-6 Luna: Model B more clearly depicts the closed laundromat, red sparkling-water vending machine, leaning umbrella, and neon-lit wet pavement, while Model A has stronger rain and a cinematic widescreen frame but leaves the laundromat less identifiable. Model B’s square framing misses the requested 16:9 format, but its overall scene matches the prompt more completely. (Second judge pass, order swapped — scores are the average of both: Model B more clearly captures the closed laundromat, red sparkling-wate | Anthropic: Claude Fable 5.1: Model A delivers the requested 16:9 cinematic f…
Exact counting
A clean overhead flat lay on white marble of EXACTLY seven identical espresso cups arranged in a single straight row, each clearly separate and individually countable, even soft lighting, 16:9.
OpenAI: GPT-6 Luna: Model B shows exactly seven matching, individually countable cups in a straight row on white marble. Model A also has seven cups, but several are cropped at the edges and the cups vary in color and contents; Model B’s main miss is the square format rather than 16:9. (Second judge pass, order swapped — scores are the average of both: Model B clearly shows exactly seven matching espresso cups in a straight, evenly spaced row on white marble, while Model A has cups clipped at the edges and visibly | Anthropic: Claude Fable 5.1: Model A shows seven cups but they are not identi…
Sleep spray botanical vector
A clean stylized vector illustration of a consumer wellness product: a frosted glass pillow-mist bottle branded "Lune Thread No. 4" surrounded by three moonflower blossoms, two sprigs of sage, and a small sleeping fox motif on the label, arranged in a balanced poster composition on a deep indigo background; elegant minimalist line art with flat shapes, restrained grain texture, and a calm dreamy mood, using pale lavender, muted silver, and soft cream accents.
OpenAI: GPT-6 Luna: Model B includes the requested three moonflowers, two sage sprigs, fox motif, and clearly rendered branding in a polished, balanced poster composition. Model A has a clean minimalist style, but appears to include only two blossoms and has less complete product labeling. (Second judge pass, order swapped — scores are the average of both: Model B closely matches the requested poster, with three moonflower blossoms, botanical sprigs, a sleeping fox, and clearly rendered “Lune Thread No. 4” branding | Anthropic: Claude Fable 5.1: Model B delivers all three moonflower blossoms,…
Matchup powered by OpenRouter.