I tested ChatGPT-5 vs Gemini 2.5 Pro with 11 AI image prompts — here's the winner
When you buy through links on our articles, Future and its syndication partners may earn a commission.
Credit: Future
ChatGPT-5 and Gemini 2.5 Pro are two of the most advanced multimodal chatbots available. Both are available with a free tier option, and can create crisp, realistic images within the chat window in seconds.
Compared to GPT-4 , OpenAI's GPT-5 takes multimodality to the next level with sharper image analysis, more natural voice and a massive 400K-token context window. The latest version is said to be safer with fewer errors and smarter at routing tasks between "chat" and "thinking" modes, making it a stronger, more versatile AI assistant.
Similarly, Gemini 2.5 Pro
is Google DeepMind
's most advanced multimodal reasoning model yet. With a massive one-million-token context window, it excels in solving complex math and science problems, surpasses competitors on benchmarks, and demonstrates exceptional coding prowess; from interactive simulations, web apps and debugging support from a single prompt, it handles images, audio, video and even entire codebases.
I just had to know how the two compared with image generation. Here's what happened when I used the same eleven prompts to create images and how the chatbots
stack up against each other.
Category 1: Mashup
Futuristic New York City
Credit: Future
Prompt: "Design a futuristic New York City street scene in 2050 where self-driving food trucks shaped like animals are serving people. Include at least three different species (like a panda truck, dolphin truck, and eagle truck), neon signage in multiple languages and rain-slick streets with reflections."
ChatGPT-5delivered on every detail of the prompt and it is clear that the image takes place in New York City, despite it being futuristic.
Gemini 2.5 Progenerated a clear and vibrant image that hit nearly every aspect of the prompt, but missed what might be the biggest one: New York City. There's no way to know where this scene is located.
Winner: ChatGPTwins for a solid aesthetic and taking every detail into account.
Unlikely jungle animals
Credit: Future
Prompt: "Create a lively jungle village where animals live like humans. A tiger is reading a newspaper on a hammock, a parrot is running a fruit stand, elephants are carrying groceries in baskets, and monkeys are playing soccer in the background. Add wooden huts, vines wrapped around them, and lanterns glowing at dusk."
ChatGPT-5took the mashup idea to new heights with creating an image that was both comical and yet realistic. I was hoping either one of the bots would go for realism over cartoon-y. ChatGPT did not disappoint but completely missed the parrot running a fruit stand.
Gemini 2.5 Proturned this prompt into a cute illustration and took into account the details but overall the image wasn't as intriguing (IMO).
Winner: Gemini winsfor taking into account every detail of the prompt. ChatGPT was very close and the image was more realistic, which I liked, but the win has to go to the one who best handled the prompt. Would love your thoughts on this one; share in the comments.
Category 2. Indoor/Outdoor
Cozy living room
Credit: Future
Prompt: "A cozy modern living room on a rainy afternoon, sunlight filtering through the window, with a golden retriever sleeping on the couch."
ChatGPT-5created an image that was realistic and cozy, but the lighting is too warm and evening-like for the prompt's specifications.
Gemini 2.5 Prodelivered a very realistic image that captures the scene with the right lighting and aesthetic, while embodying the prompt completely.
Winner: Geminidid a better job at accurately capturing the "rainy afternoon with sunlight filtering through" lighting condition, rendered the golden retriever more clearly and created a scene that better matches the modern, cozy aesthetic described.
Street market
Credit: Future
Prompt: "A street food market in Bangkok at night, neon lights reflecting off wet pavement, bustling crowds, and steam rising from food stalls."
ChatGPT-5crafted vivid and colorful reflections that immerse the viewer in the atmospheric scene. The composition is cinematically appealing, and the balance between all the elements — such as steam, crowd, neon signs and wet reflection — feel harmonious.
Gemini 2.5 Prodelivered a strong urban atmosphere with the busy street scene and included excellent signage with authentic Thai text. The wet pavement reflections are good, but could be more pronounced.
Winner: ChatGPTdid a superior job with this prompt. While both captured the essential elements, ChatGPT excelled particularly in the "neon lights reflecting off wet pavement" aspect, which is a key detail in the prompt.
Category 3: Creativity and surrealism
Futuristic city
Credit: Future
Prompt: "A futuristic city floating above the ocean, powered by giant glowing energy crystals, with flying cars zipping between the towers."
ChatGPT-5delivered an image with a more fantastical and visually striking interpretation with giant glowing energy crystals. While the crystals look as though they are powering the city, the overall image lacks the depth and realistic nuance of Gemini.
Gemini 2.5 Procreated a clean image with sleek, futuristic architecture consisting of tall spires and modern buildings. It is more realistic with a good sense of scale and urban planning depicted.
Winner: Geminiwins for creating both a fantastical and realistic image that unmistakably showcases the city's power source of brilliantly glowing crystals on the building. Overall, the sci-fi concept better matches the goal of the prompt.
Out-of-this-world tea party
Credit: Future
Prompt: "A tea party on the moon with astronauts and alien creatures, pastel-colored plates, and a view of Earth in the background."
ChatGPT-5crafted a charming illustration that is whimsical and storybook-like. The pastel color scheme throughout the entire image is pleasant and intriguing.
Gemini 2.5 Prodelivered a photorealistic approach with proper lunar surface texture and astronauts looking appropriately suited for space exploration. The detailed view of Earth in the background is a nice addition to the details.
Winner: Geminiwins for best executing the prompt with a beautiful, realistic space scene and adorable "alien creatures" as well as the extra details within the prompt.
Category 4: People & portraits
Day in Paris
Credit: Future
Prompt: "A candid photo of a middle-aged woman laughing at an outdoor café in Paris, with the Eiffel Tower blurred in the background."
ChatGPTcreated a less authentic image as it feels much more posed and professional-looking, rather than capturing candid laughter. The Eiffel Tower background lacks the proper depth and blur effect mentioned in the prompt.
Gemini 2.5 Proproperly blurred the Eiffel Tower in the background creating good depth of field while focusing on the natural, genuine laughter of the woman in the image. The overall composition feels like an actual candid street photograph.
Winner: Geminiwins for capturing the "candid" nature of the prompt significantly better.
Soccer practice
Credit: Future
Prompt: "A group of five kids playing soccer in a grassy park at sunset, motion blur capturing movement, with diverse ethnicities represented."
ChatGPT-5delivered a more static composition and the sunset lighting feels less dramatic. The children appear more posed than in motion.
Gemini 2.5 Proshowed clear motion blur effect with children running with a visible soccer ball and a good sense of movement and energy. The golden hour sunset lighting is realistic as are the diverse group of children playing.
Winner: Geminiwins for handling the prompt the best and creating a genuine feeling of kids running and playing with effective motion blur that captures the energy of children in motion.
Category 5: Style flexibility
Art style and texture realism
Credit: Future
Prompt: "An Impressionist oil painting of a quiet canal in Venice at sunrise, inspired by Claude Monet's color palette and brushstrokes."
ChatGPT-5truly captured an Impressionist technique with visible, expressive brushstrokes throughout.
Gemini 2.5 Procrafted a beautiful painting, but the image is more photorealistic than in a true Impressionist style
Winner: ChatGPTwins for executing the prompt significantly better. While Gemini created a pleasant Venice scene, it missed the core requirement of Impressionist style, which ChatGPT delivered. The chatbot's image captures the essence of how Monet would approach a Venice sunrise, prioritizing the impression of light and color over literal architectural detail.
Anime-style poster design
Credit: Future
Prompt: "A Japanese anime-style poster featuring a teenage girl standing on a rooftop at dusk, cherry blossoms falling around her."
ChatGPTemphasized the character rather than environmental details, and the image was more anime than poster.
Gemini 2.5 Procreated a polished, professional poster design with Japanese text and title treatment.
Winner: Geminihandled this prompt more effectively. The keyword was "poster." Gemini created an actual poster design complete with Japanese text, proper composition and professional layout that you'd see for an anime film or series.
Category 6: Object accuracy & branding
9. Marketing style
Credit: Future
Prompt: "A top-down flat-lay photo of the latest iPhone on a wooden desk, next to AirPods and a cup of coffee, shot in a clean Apple-style aesthetic."
ChatGPT-5was correct with its top-down angle and wooden surface, but missed the mark with a more casual, lived-in feeling rather than pristine Apple aesthetic
Gemini 2.5 Protruly nailed the aesthetic with a perfect top-down shot, clean, minimalist composition with ample white space. The iPhone with modern design and colorful wallpaper offered exceptional detail.
Winner: Geminiexecuted the "clean Apple-style aesthetic" requirement significantly better. The composition has a pristine, minimalist quality that defines Apple's marketing photography. The difference in attention to the aesthetic requirement is quite clear between the two approaches.
Overall winner: Gemini 2.5 Pro
After testing nine diverse prompts across five categories, Gemini 2.5 Pro emerges as the stronger image generator, winning six out of nine comparisons. Google's model consistently excelled at photorealism and technical accuracy, particularly when prompts contained specific requirements like lighting conditions, motion blur or aesthetic styles. Gemini demonstrated superior attention to detail in scenarios ranging from candid street photography to Apple-style product shots.
ChatGPT-5 showed its strengths in artistic interpretation and atmospheric effects, delivering standout results when creative expression mattered more than technical precision.
The results reveal distinct philosophies: Gemini tends toward technical precision and literal prompt adherence, while ChatGPT leans into artistic interpretation and visual impact. For users prioritizing accuracy and detailed prompt execution, Gemini 2.5 Pro proves more reliable. However, ChatGPT-5 remains valuable when seeking creative flair and atmospheric storytelling.
Both models represent significant advances in AI image generation, offering near-instant, high-quality results that would have required professional photographers or artists just years ago. The choice between them ultimately depends on whether you value technical precision or creative interpretation more highly in your visual content needs.
Follow
Tom's Guide on Google News
to get our up-to-date news, how-tos, and reviews in your feeds. Make sure to click the Follow button.
