Image Battle
Compare AI Image Generators for your use-case
Head to head

GPT Image 1.5 vs MAI-Image-2.6

MAI-Image-2.6 wins 59% of 84 direct comparisons against GPT Image 1.5.

Both models received the same prompts. Wins and losses count only images judged together in the same call, the same comparisons the leaderboard Elo is fitted from.

GPT Image 1.5
OpenAI
1094 Elo ±45
Rank #13 of 42 · median 37s per image · 11 refused
28–13–43
wins – ties – losses for GPT Image 1.5
84 direct comparisons
Elo expectation: GPT Image 1.5 wins 43%
MAI-Image-2.6
Microsoft
1145 Elo ±59
Rank #10 of 42 · median 69s per image · 15 refused

By category

CategoryGPT Image 1.5MAI-Image-2.6Direct record
Anime & Cartoon Style 1128 1297 0–2–4
Architecture & Interiors 1160 1138 3–3–3
Complex Scenes 1070 1122 3–1–5
Ghibli style 1204 1490 0–0–1
Graphic Design 930 1048 3–1–6
Hands & Anatomy 1114 1115 4–2–3
Photorealistic People & Portraits 1108 1115 5–1–4
Surreal & Creative Prompts 1166 1289 2–1–7
Text in Images 1003 1090 3–0–7
Ultra Hard 1286 1273 5–2–3

By criterion

CriterionGPT Image 1.5MAI-Image-2.6
Prompt adherence 1135 1110
Technical quality 1034 1102
Artistic merit 1079 1157
Detail execution 1072 1125
Realism & coherence 1006 1115

Side by side

Anime & Cartoon Style An anime-style portrait of a samurai standing under pink cherry blossoms.
An anime-style portrait of a samurai standing under pink cherry blossoms. - GPT Image 1.5
GPT Image 1.5 7 / 10 Rank 10 of 13: competent and attractive, with clear anime and samurai cues. Excessive surface busyness and less controlled fabric rendering prevent a higher placement.
An anime-style portrait of a samurai standing under pink cherry blossoms. - MAI-Image-2.6
MAI-Image-2.6 9 / 10 Rank 1 of 13: the strongest combination of anime styling, sophisticated color, and controlled composition. Minor simplification in distant details keeps it below a flawless overall score.
Architecture & Interiors Cross-section rendering of a multi-level underground bunker with different functional areas, structural support systems, ventilation, and power generation facilities visible.
Cross-section rendering of a multi-level underground bunker with different functional areas, structural support systems, ventilation, and power generation facilities visible. - GPT Image 1.5
GPT Image 1.5 6 / 10 An attractive and broadly relevant illustration, but its engineering organization is weaker than its visual presentation. Ambiguous circulation and loosely resolved machinery materially reduce its value as a technical cross-section.
Cross-section rendering of a multi-level underground bunker with different functional areas, structural support systems, ventilation, and power generation facilities visible. - MAI-Image-2.6
MAI-Image-2.6 8 / 10 This is one of the most visually accomplished and informative submissions. Its promotional engineering claims and schematic auxiliary systems make it less coherent than the two strongest architectural renderings.
Complex Scenes A bustling market scene with dozens of people buying and selling goods.
A bustling market scene with dozens of people buying and selling goods. - GPT Image 1.5
GPT Image 1.5 6 / 10 Vivid and highly faithful to the prompt, but weaker on the close-up interaction that should anchor the scene. The ambiguous weighing apparatus and soft hand structure outweigh its attractive atmosphere.
A bustling market scene with dozens of people buying and selling goods. - MAI-Image-2.6
MAI-Image-2.6 9 / 10 Ranks second, closely matching the leading image in prompt coverage and environmental execution. Slightly less precise foreground hand contacts are the main reason it falls behind the first-ranked scene.
Ghibli style A child and elder tending oversized vegetables in a garden. Reproduce the artistic style of Studio Ghibli, with their distinctive character proportions and detailed plant illustrations.
A child and elder tending oversized vegetables in a garden. Reproduce the artistic style of Studio Ghibli, with their distinctive character proportions and detailed plant illustrations. - GPT Image 1.5
GPT Image 1.5 7 / 10 Ranked 7th of 13: this is a lively and richly populated interpretation with clear gardening activity. Its crowded staging, coarse highlights, and minor botanical and contact ambiguities keep it below the more controlled illustrations.
A child and elder tending oversized vegetables in a garden. Reproduce the artistic style of Studio Ghibli, with their distinctive character proportions and detailed plant illustrations. - MAI-Image-2.6
MAI-Image-2.6 9 / 10 Ranked 1st of 13: this offers the best balance of faithful character design, convincing gardening gestures, and richly differentiated botany. Its consistency across anatomy, materials, composition, and emotional tone makes it the strongest complete interpretation.
Graphic Design Set of 5 consistent flat design icons for a mobile banking app: 'Home', 'Transfer', 'Account', 'Scan QR', 'Settings'. Monochrome dark blue.
Set of 5 consistent flat design icons for a mobile banking app: 'Home', 'Transfer', 'Account', 'Scan QR', 'Settings'. Monochrome dark blue. - GPT Image 1.5
GPT Image 1.5 4 / 10 Complete concept coverage is undermined by severe presentation problems. Poor contrast and non-flat rendering outweigh the correct labels and recognizable symbols.
Set of 5 consistent flat design icons for a mobile banking app: 'Home', 'Transfer', 'Account', 'Scan QR', 'Settings'. Monochrome dark blue. - MAI-Image-2.6
MAI-Image-2.6 9 / 10 This joins the leading result through exceptional stylistic cohesion and clean flat execution. Its only meaningful limitation is the highly simplified QR interior, which remains recognizable at icon scale.
Hands & Anatomy A person typing on a laptop keyboard, showing both hands on the keys.
A person typing on a laptop keyboard, showing both hands on the keys. - GPT Image 1.5
GPT Image 1.5 9 / 10 Ranked 1st of 13, this image best addresses the core test of realistic anatomy and interaction. Small device-label imperfections prevent a flawless score, but its hands and contact are exceptionally convincing.
A person typing on a laptop keyboard, showing both hands on the keys. - MAI-Image-2.6
MAI-Image-2.6 3 / 10 Ranked 13th of 13, this is the weakest result because of its critical extra-hand problem. Attractive lighting and respectable textures cannot rescue the anatomical incoherence and distorted laptop.
Photorealistic People & Portraits A 4K photo of an old fisherman with a weathered face and a pipe, grinning.
A 4K photo of an old fisherman with a weathered face and a pipe, grinning. - GPT Image 1.5
GPT Image 1.5 6 / 10 Ranks seventh for memorable expression and strong subject coverage. The visibly exaggerated skin treatment is a substantial limitation for a prompt specifically testing photographic age realism.
A 4K photo of an old fisherman with a weathered face and a pipe, grinning. - MAI-Image-2.6
MAI-Image-2.6 9 / 10 Rank 2 of 13: one of the most believable and characterful photographs in the set. Its slightly softer finish is less immediately striking than the winner, but its natural aging treatment is excellent.
Surreal & Creative Prompts The Mona Lisa reimagined as a futuristic android, rendered in a digital art style that preserves Da Vinci's composition while incorporating sleek cyberpunk aesthetics and technological elements.
The Mona Lisa reimagined as a futuristic android, rendered in a digital art style that preserves Da Vinci's composition while incorporating sleek cyberpunk aesthetics and technological elements. - GPT Image 1.5
GPT Image 1.5 6 / 10 Ranks eighth because it is a competent futuristic portrait but only a partial preservation of the specified composition. The generic combat armor and interrupted hand framing are substantial prompt-level weaknesses.
The Mona Lisa reimagined as a futuristic android, rendered in a digital art style that preserves Da Vinci's composition while incorporating sleek cyberpunk aesthetics and technological elements. - MAI-Image-2.6
MAI-Image-2.6 9 / 10 Ranks first for its exceptionally cohesive balance of recognizable composition, inventive android design, and sophisticated environmental storytelling. It avoids the generic armor treatment and disconnected technological overlays seen in weaker entries.
Text in Images A wrinkled white T-shirt with "Carpe Diem" in bold serif font and "Seize The Day" in smaller script below it, modeled on a mannequin, professional product photography with soft-box lighting.
A wrinkled white T-shirt with "Carpe Diem" in bold serif font and "Seize The Day" in smaller script below it, modeled on a mannequin, professional product photography with soft-box lighting. - GPT Image 1.5
GPT Image 1.5 8 / 10 Ranked 2nd of 13, this is an excellent and highly faithful product image. Its slightly pristine print treatment gives the top-ranked entry an advantage in fabric–text integration.
A wrinkled white T-shirt with "Carpe Diem" in bold serif font and "Seize The Day" in smaller script below it, modeled on a mannequin, professional product photography with soft-box lighting. - MAI-Image-2.6
MAI-Image-2.6 4 / 10 Ranked 13th of 13, this has competent photographic qualities but fails the central wording requirement. Because clear and accurate text is a primary purpose of the test, the missing 'Diem' outweighs the otherwise believable fabric.
Ultra Hard Photorealistic image of a robot painting a realistic self-portrait (i.e. the robot) on canvas, mimicking Van Gogh’s art style; clear, realistic metallic textures and painting details visible
Photorealistic image of a robot painting a realistic self-portrait (i.e. the robot) on canvas, mimicking Van Gogh’s art style; clear, realistic metallic textures and painting details visible - GPT Image 1.5
GPT Image 1.5 5 / 10 More imaginative and self-portrait-aware than the human-portrait variants, but it fails the photorealistic outer-scene requirement. Its high painterly merit cannot compensate for that central stylistic mismatch.
Photorealistic image of a robot painting a realistic self-portrait (i.e. the robot) on canvas, mimicking Van Gogh’s art style; clear, realistic metallic textures and painting details visible - MAI-Image-2.6
MAI-Image-2.6 9 / 10 A close second to 4430, with outstanding material realism and a richly integrated studio setting. Its busier staging and slightly less precise artist-to-portrait likeness keep it below a perfect result.

More comparisons