Compare AI Image Models

Every model, every prompt. Click a column header to dig into one prompt; click a row label to see what a model can do across all of them.

Model ↓ / Prompt →	Fairy-Tale Poster	Isometric Cafe Cutaway	Phone-Screen Mirror Selfie	Art Nouveau Travel Poster	Same Character, Four Panels	Only in the Reflection	A Cat With a Horse's Body	A Cat the Size of a Horse	Cyan, Magenta, Yellow: Three Spotlights	Three Spotlights, One Wall	Three Lamps, Three Blocks	Three Lamps, One Hand	Railing Shadow on a Staircase	Bioluminescent Deep-Sea Parade	Two People, Long Shadows (Overhead)	Two Arrows, One Glass of Water	A Bowl of Wrong-Coloured Fruit	Honey-Drip Typography	Pencil in a Glass of Water	Backyard Cryptid on a Trail Cam	Empty Desert Highway at Sunset	Pixel-Art RPG Shopkeeper, 32-Bit Era	Watercolor Botanical — Hummingbird and Trumpet Vine	Bizarre Plush Toy Web Catalog	Stack of Pancakes with Melting Butter	Wes Anderson Hotel Lobby	1960s Pulp Paperback Cover	Hands Holding a Coffee Cup	Tokyo Alley After Rain	Vintage 1980s Cereal Box	Windows 7 Desktop, Early 2010s	Brutalist Cafe Sign at Golden Hour	Dramatic Mountain Silhouette
ChatGPT (GPT Image 2) OpenAI ChatGPT prompts
Nano Banana 2 (Gemini) Google Gemini 2.5 Flash Image (Nano Banana) - Advanced image generation and editing
Flux Kontext Pro Black Forest Labs' Flux Kontext Pro — image-to-image specialist with strong identity preservation. Routed via AIML API.																					Run this →		Run this →		Run this →			Run this →		Run this →	Run this →
Nano Banana Pro Google's flagship Gemini 3 Pro Image model — strongest prompt adherence and text rendering in the Gemini family. Hosted via AIML API.	—	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	—
FLUX.2 Max Black Forest Labs' FLUX.2 generation — Max tier targets editorial-grade composition and detail. Hosted via AIML API.	—	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →		Run this →		Run this →	—
Grok Imagine Pro xAI's image model — distinctive stylistic signature, strong on cinematic and photoreal scenes. Hosted via AIML API.	—	Run this →	Run this →	Run this →	Run this →		Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →	Run this →		Run this →		Run this →													—
Seedream 5 Lite ByteDance's Seedream 5 — Lite tier, fast and competitive on photoreal and Asian-style aesthetic. Hosted via AIML API.	—	Run this →	Run this →																		Run this →				Run this →		Run this →	Run this →		Run this →	Run this →	Run this →	—
Qwen Image 2.0 Pro Alibaba's flagship Qwen Image 2.0 Pro — strong text rendering and multilingual prompts. Hosted via AIML API.	—																				Run this →		Run this →					Run this →	Run this →	Run this →		Run this →	—
Nano Banana 2 Lite Google's lightweight Gemini Flash image model — the faster, cheaper sibling of Nano Banana 2, tuned for speed and volume over maximum fidelity. Hosted via AIML API.	—

9 models × 37 prompts. Click any output to zoom in. Empty cells haven't been benchmarked yet — click Run this to try the prompt yourself.