Head to head: Kling Image vs Ideogram V4.5 Edit
Kling Image vs Ideogram V4.5 Edit
By Ryan Merket · Published
This image-model matchup spans stylized art, exact text and object placement, restricted palettes, counting, and food photography. The key question is whether stronger prompt fidelity across varied briefs outweighs a couple of precise wins.
Ideogram V4.5 Edit leads the saved task sample with an aggregate score of **65.0** to Kling Image’s **55.4**, and takes six task wins to Kling’s two, with no ties. That is a **clear result** in this comparison—not a claim of statistical significance or a guarantee beyond these prompts. Ideogram’s edge comes from repeatedly satisfying the brief’s defining constraints: it captured the cresting-wave ukiyo-e scene, followed the object-placement instructions, stayed closer to the restricted palette, and showed exactly 11 citrus tarts. It also better delivered the tight macro framing and linen texture for the baklava, and made the pomegranate-soda opening action more legible and cinematic. Kling had two notable, specific wins. Its neon dumpling poster rendered both lines of text exactly and made the bamboo steamer a stronger centerpiece; its overhead espresso image showed seven clearly separated cups without the extra saucers and spoons in Ideogram’s version. Those are meaningful advantages on text legibility and clean counting, but they did not outweigh Ideogram’s broader run of constraint-following wins. **Verdict: Ideogram V4.5 Edit wins this head-to-head on the saved task sample.** The assessment applies only to these tasks; it should not be read as a general ranking of the models.
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 captures the ukiyo-e tradition with a towering crest, stylized foam, a small fishing boat, and a dawn palette; Model A has appealing woodblock-like linework but its boat feels more like it is floating on choppy water than riding a large cresting wave. Model B also has the more dramatic, cohesive composition. (Second judge pass, order swapped — scores are the average of both: Model B more closely matches the prompt with a dramatic cresting wave, a fishing boat and crew, dawn su | Anthropic: Claude Fable 5.1: Model B delivers a true large cresting wave wit…
Neon dumpling poster text
A 1980s retro print style street-food poster in a rainy night market, centered on a glowing teal-and-magenta neon sign above a bamboo steamer, with the sign text rendered clearly and exactly as: "LUNA DUMPLING BAR" and a smaller line beneath reading "OPEN TIL 2AM"; bold high-contrast typography, crisp legibility, glossy reflections on wet pavement, dramatic rim lighting, symmetrical composition, 16:9.
OpenAI: GPT-6 Luna: Model A delivers the exact, exceptionally legible sign text and a prominent bamboo steamer amid rain, neon, and wet reflections, though its rendering is less distinctly retro-print. Model B has a stronger vintage print mood and a dramatic market alley, but the steamer is small and closed, making the food-poster centerpiece less effective. (Second judge pass, order swapped — scores are the average of both: Model B has a stronger 1980s retro-print atmosphere and an appealingly balanced night-marke | Anthropic: Claude Fable 5.1: Model A renders both text lines exactly and in…
Attribute binding
A photorealistic tabletop scene: a matte green cube to the LEFT of a glossy red sphere, a small blue cylinder BEHIND both, and a single yellow rubber duck sitting ON TOP of the cube. Colors must stay on the correct objects. Soft studio light, 16:9.
OpenAI: GPT-6 Luna: Model B more clearly places the duck on the green cube, the red sphere to its right, and a blue cylinder behind them; Model A adds a conspicuous blue sphere atop the cylinder. Model A has more convincing photographic texture, while Model B is cleaner but somewhat more CGI-like. (Second judge pass, order swapped — scores are the average of both: Model B follows the object and placement constraints more precisely, with the blue cylinder behind both objects and the duck on top of the cube. Model A | Anthropic: Claude Fable 5.1: Model A delivers a convincingly photorealistic s…
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 B has a clearer, more balanced campsite-at-dusk composition and a cohesive flat-vector style, while Model A’s oversized foreground tent and bright cyan/red accents make the scene feel crowded and stray from the restricted palette. Model B also includes some white and lighter teal tones beyond the four specified colors, but adheres more closely overall. (Second judge pass, order swapped — scores are the average of both: Model B better follows the four-color restriction and presents a clear, | Anthropic: Claude Fable 5.1: Model A breaks the four-color constraint with b…
Count the citrus tarts
A clean editorial overhead shot of a patisserie display on pale stone, showing exactly 11 distinct mini citrus tarts arranged in loose staggered rows—each tart fully visible and individually countable—with varied toppings of candied kumquat, lime zest curls, blood orange slices, and tiny mint leaves; soft morning window light, minimal props, precise composition, high realism, 16:9.
OpenAI: GPT-6 Luna: Model B shows exactly 11 individually countable tarts, while Model A appears to show only 10 and crops a tart at the right edge. Model A is attractive, but Model B better fulfills the count and clear-display requirements. (Second judge pass, order swapped — scores are the average of both: Model B shows exactly 11 distinct, fully visible tarts in loose rows and closely matches the minimal pale-stone editorial brief. Model A is attractive and richly detailed, but appears to show only 10 tarts, inc | Anthropic: Claude Fable 5.1: Model B shows exactly 11 distinct, fully visibl…
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 A shows exactly seven clearly separated cups in a straight overhead row on white marble, with no distracting extras. Model B also has seven countable cups, but the added saucers and spoons depart from the prompt and make the cups less distinct. (Second judge pass, order swapped — scores are the average of both: Model A presents exactly seven clearly separated, matching cups in a straight overhead row, with more consistent spacing and orientation. Model B also meets the count and has softer | Anthropic: Claude Fable 5.1: Model A shows exactly seven identical espresso…
Baklava layers macro
An extreme macro studio photograph of a diamond-cut pistachio baklava on a dark ceramic plate, showing razor-sharp flaky phyllo layers, sticky amber syrup threads, crushed pistachio granules, tiny bubbles in the glaze, and a visible linen napkin weave beside it; cinematic side lighting, shallow depth of field with the front edge tack sharp, luxurious fine texture without smudging, 16:9.
OpenAI: GPT-6 Luna: Model B better captures the extreme macro feel and includes the clearly visible linen weave, while retaining glossy syrup and crisp phyllo detail. Model A has sharper detail across more of the pastry and a stronger view of the dark plate, but misses the requested napkin texture and feels less tightly cropped. (Second judge pass, order swapped — scores are the average of both: Model B better captures the requested extreme macro, with crisp flaky layers, glossy syrup, pistachio pieces, a dark plat | Anthropic: Claude Fable 5.1: Model B better fulfills the extreme-macro brief…
Pomegranate soda burst
A hyper-real food photography scene of a tattooed bartender snapping a chilled bottle of sparkling pomegranate soda open mid-swing, cap flying, crimson spray arcing through the air, ice shards and droplets frozen sharply, bar towel whipping from the motion, dynamic diagonal composition, dramatic backlighting and hard sidelight, intense sense of energy and movement, 16:9.
OpenAI: GPT-6 Luna: Model B more clearly captures the bottle opening mid-swing, with a flying cap, suspended ice, crimson spray, and a whipping towel in a cinematic diagonal composition. Model A has striking detail and a broad spray, but the opening action is less legible and the towel and droplets feel less naturally integrated. (Second judge pass, order swapped — scores are the average of both: Model B more completely conveys the opening action, with a visible soda burst, airborne ice, tattoos, and a whipping tow | Anthropic: Claude Fable 5.1: Model B delivers a far more dynamic, dramatical…
Matchup powered by OpenRouter.