“I would put Google's Nano Banana and OpenAI's GPT Image at the top of your shortlist.”
“Particularly strong at understanding complicated scene descriptions.”
“Good at preserving object attributes, positions, and relationships.”
“Useful for multi-character scenes and complex environments.”
“Excellent conversational editing capabilities.”
“Excellent photorealistic image generation.”
“Good at handling detailed visual descriptions.”
“Offers flexibility for advanced workflows and model configurations.”
“Can require more experimentation to preserve every instruction.”
“Best artistic aesthetics”
“Exceptional atmosphere, lighting, composition and artistic detail.”
“Excellent at producing visually cohesive images.”
“Can reinterpret instructions in favour of aesthetics.”
“Not always reliable with exact object counts and precise spatial relationships.”
“Particularly good at text inside images.”
“Useful for posters, signs, labels and designed compositions.”
“Can struggle with prompts containing many unrelated scene requirements.”
“Benchmark Lab detailed-prompt testing | ChatGPT Images, Gemini”
“I would put Google's Nano Banana and OpenAI's GPT Image at the top of your shortlist.”
“Best for instruction following”
“Very good at interpreting lengthy natural-language instructions.”
“Strong at exact object placement and scene composition.”
“Particularly useful when prompts contain many specific constraints.”
“Can make targeted changes through conversational editing.”