Head to head: Bagel vs ImagineArt 1.5 Pro Preview

Bagel didn’t just lose this comparison — it got swept. The aggregate score gap, 69.8 to 48.7, is huge, and the task count is even harsher: ImagineArt 1.5 Pro Preview won all 8 of 8 tasks, with 100% confidence in the overall verdict. That’s not a vibes-based edge; it’s a decisive result backed by complete category coverage.

What stands out is how consistent the pattern was. Bagel occasionally looked more restrained or cleaner on a narrow reading of a requirement — especially on strict-count or limited-palette discipline — but then failed the actual assignment. In Bound Colors Relic Table, it kept a tidier still-life arrangement but swapped out key objects and missed material cues. In Restricted palette, it respected the color limitation better, yet barely produced a campsite. That happened over and over: Bagel could sometimes follow one rule while dropping the scene, subject, or core concept.

ImagineArt, by contrast, kept winning on the thing that matters most: delivering the requested image. It was more faithful to the named style in the ukiyo-e task, far more complete on dense prompt-binding in Mosswright Console Detail and Sableport Art Deco Poster, and dramatically better at action and subject fidelity in Kite-Lancer Midair Turn, where Bagel rendered the wrong creature entirely. It also handled practical failure points better — more believable hands in the bracelet photo, and much stronger negation control in the reading-nook prompt, where Bagel simply ignored the “no plants, no lamps, no artwork” constraints.

The most telling detail is that even where ImagineArt had misses — an extra object in the relic-table scene, some shading leakage in the restricted-palette image, the wrong aspect ratio in the hands task — it still produced images that were recognizably on brief and compositionally convincing. Bagel’s misses were more fundamental: wrong objects, wrong subject, missing scene elements, weaker environment building, and repeated prompt-binding failures. Clean execution is nice; correct execution wins comparisons.

Final call: ImagineArt 1.5 Pro Preview is the clear winner. Bagel showed occasional constraint discipline, but ImagineArt was better at actually satisfying prompts, and it earned a total sweep with decisive statistical confidence.

How they were tested

We ran 8 fresh image tasks, generated on the fly for this matchup so neither model could prepare in advance, and had gpt-5.4 score each one. To cancel position bias, every task was judged twice — once in each presentation order — and every number reported here, including the headline totals, is the average of both passes. Bagel scored 48.6 to ImagineArt 1.5 Pro Preview's 69.8.

1. Bound Colors Relic Table

On a black basalt ritual table in a sci-fi archive vault, arrange exactly five distinct objects with strict attribute binding: a matte crimson ceramic orb, a translucent teal glass dagger, a brushed gold metal cube, a deep violet velvet mask, and a pale jade stone key; no color swapping, no material confusion, each object clearly separated and fully visible; depicted in polished speculative still-life illustration with precise studio rim lighting, soft reflections, symmetrical front-facing composition, 16:9.

Winner: ImagineArt 1.5 Pro Preview — Model A has a cleaner front-facing still-life setup, but it fails key prompt bindings by replacing the pale jade stone key with a small pouch-like object and making the mask appear hard rather than velvet. Model B includes all five required object types with stronger sci-fi archive context and polished lighting, though it adds an extra teal sword-like object and is less symmetrical/front-facing than requested. (Second judge pass, order swapped — scores are the average of both: Model B has a richer sci-fi archive setting and polished rendering, but it violates the prompt by including an extra teal blade-like object and using a glossy orb rather than a clearly matte ceramic one. Model A adheres more closely to the exact count, separation, and front-facing still-life arrangement, though it misses the archive-vault environment and some specified material cues such as velvet and stone.)

2. 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.

Winner: ImagineArt 1.5 Pro Preview — Model A follows the restricted four-color flat-vector requirement much more closely, but it fails the campsite-at-dusk brief because there is no visible camp setup and the composition feels sparse. Model B delivers a strong mountain campsite scene with superior composition and visual appeal, but it violates the flat-vector and strict four-color palette constraints with shading, texture, and likely off-palette tones. (Second judge pass, order swapped — scores are the average of both: Model B clearly depicts a mountain campsite at dusk in a strong flat-vector style with a balanced 16:9 composition, though it appears to introduce subtle texture/gradient-like shading and tiny text artifacts that weaken strict palette adherence. Model A uses a simpler restricted palette cleanly, but it fails the campsite-at-dusk brief by omitting tents and campfire and feels underdeveloped compositionally.)

3. 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.

Winner: ImagineArt 1.5 Pro Preview — Model B better captures the ukiyo-e woodblock print tradition through more authentic wave stylization, flatter color handling, and a composition that feels closer to the referenced historical aesthetic. Model A is attractive and clear, but its boat design and cleaner, more modern illustration feel make it less faithful to the prompt’s muted palette and traditional woodblock character. (Second judge pass, order swapped — scores are the average of both: Model B is much more faithful to the ukiyo-e woodblock print prompt, with flat color planes, bold outlines, stylized foam, and a muted indigo-and-cream palette; its composition also strongly evokes the intended traditional print. Model A has a clear and appealing composition, but the modern fishing boat design and cleaner digital look make it less faithful to the named art style and palette.)

4. Mosswright Console Detail

A close three-quarter view of a dwarven mosswright repairing an ancient astro-navigation console inside a cliffside observatory, every surface packed with crisp micro-detail: frayed indigo wool cuffs with visible weave, frost-tipped fox fur lining on the hood, beadlike moss and tiny fern fronds growing between engraved brass gears, pitted iron rivets, etched crystal lenses, and a tray of miniature silver tools; rendered in hyper-detailed fantasy realism with cool dawn light slanting through a hexagonal window, shallow atmospheric haze, razor-sharp focus on textures, cinematic composition, 16:9.

Winner: ImagineArt 1.5 Pro Preview — Model B better matches the prompt with a clearly ancient astro-navigation console packed with engraved gears, crystal lenses, moss, fern fronds, and a tray of miniature silver tools, while also using the hexagonal window and dawn beams more effectively. Model A is attractive and sharp, but it underdelivers on the console complexity and micro-detail specificity, and the observatory/astro-navigation aspect feels much weaker. (Second judge pass, order swapped — scores are the average of both: Model B matches the prompt more completely with a clear close three-quarter view, richly detailed astro-navigation console, visible moss and fern growth among engraved gears, crystal lenses, and a tray of miniature silver tools, all rendered with strong cinematic dawn lighting and texture focus. Model A is attractive and atmospheric, but the console is much simpler, the micro-detail density is lower, and several specified elements like engraved brass gears and the elaborate observatory machinery are less fully realized.)

5. Sableport Art Deco Poster

A glamorous night-time travel poster for the floating metropolis of Sableport, rendered faithfully in 1920s Art Deco style: stepped skyscrapers with sunburst crowns, streamlined sky-ferries docking at a crescent aerial terminal, elegant figures in geometric eveningwear on a balcony, bold zigzag borders, lacquered black, copper, cream, and emerald palette, stylized clouds and radiating beams, flat graphic shapes with luxurious symmetry, like a premium railway advertisement, 16:9.

Winner: ImagineArt 1.5 Pro Preview — Model B adheres much more closely to the prompt, with a convincing 1920s Art Deco travel-poster look, aerial terminal, multiple sky-ferries, elegant balcony figures, zigzag border, and the specified lacquered black/copper/cream/emerald palette. Model A has some Deco skyscraper styling and poster framing, but it misses key floating-metropolis and aerial-terminal elements, reads more like a waterfront rail scene, and its text rendering is noticeably garbled. (Second judge pass, order swapped — scores are the average of both: Model B matches the prompt far more closely with a richly symmetrical 1920s Art Deco travel-poster look, floating metropolis setting, aerial terminal activity, balcony figures, zigzag border, and the specified lacquered black/copper/cream/emerald palette. Model A has a strong Deco poster feel but misses key prompt elements like the floating aerial-terminal metropolis and sky-ferries, and its text rendering is noticeably incorrect.)

6. Kite-Lancer Midair Turn

A young storm-lancer on a biomechanical kite-dragon whips through a sudden midair hairpin turn above the salt spires of Orun Vale, one hand gripping crackling reins while the other lowers a barbed lance, cape snapping sideways, sand and rain droplets trailing in curved streaks, wing membranes flexed under strain, the mount’s tail carving an S-shape through the air; painted as a high-energy cinematic fantasy action scene with dramatic foreshortening, low-angle composition, split lighting from sunset and lightning, intense sense of motion, 16:9.

Winner: ImagineArt 1.5 Pro Preview — Model B matches the prompt far more closely with a biomechanical kite-dragon, barbed lance, violent midair turn, split sunset-and-lightning lighting, and strong motion cues; Model A is visually appealing but depicts more of a winged horse rider than a storm-lancer on a biomechanical kite-dragon and misses much of the specified action. Model B also has the more dynamic low-angle cinematic composition and stronger overall energy, despite some clutter and minor anatomical/mechanical complexity issues. (Second judge pass, order swapped — scores are the average of both: Model B strongly matches the prompt with a biomechanical kite-dragon, dramatic low-angle action, lightning/sunset split lighting, curved motion trails, and a convincing midair turn; Model A is visually clean but misses the core subject by depicting a winged horse rather than a biomechanical kite-dragon and lacks the same cinematic motion and storm intensity.)

7. 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.

Winner: ImagineArt 1.5 Pro Preview — Model B better matches the prompt with a more natural candid photo feel, clearly visible hands, and a more plausible grip while tying the bracelet threads. Model A is aesthetically pleasing and clean, but it feels more staged, has less convincing bracelet-tying action, and the hand anatomy/grip is slightly less natural. (Second judge pass, order swapped — scores are the average of both: Model B better matches the prompt with a more candid, natural photo of bracelet-tying, both hands clearly visible, believable finger anatomy, and realistic daylight; its only notable miss is the square framing rather than 16:9. Model A is aesthetically pleasing and cleaner, but feels more staged, shows a less convincing tying action with simplified thread interaction, and also misses the requested 16:9 aspect ratio.)

8. 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.

Winner: ImagineArt 1.5 Pro Preview — Model A captures a cozy reading nook with warm light, but it clearly violates the negation constraints by including multiple plants and framed wall art. Model B adheres much better to the prompt with an armchair, books, tea, window, and no visible plants, lamps, or artwork, while also offering strong composition and appealing warm afternoon atmosphere. (Second judge pass, order swapped — scores are the average of both: Model B adheres closely to the prompt with a cozy armchair, books, tea by a window, warm afternoon light, and crucially no plants, lamps, or artwork visible. Model A is attractive and well-composed, but it clearly violates the negation constraints by including multiple plants and framed wall art.)

See every prompt and the full side-by-side outputs in the interactive Head-to-Head.