The new GPT Image 1.5 vs Nano Banana Pro: Who Has the Better Image Model?

The AI image generation landscape just got more competitive. OpenAI recently launched ChatGPT Image 1.5 (also known as GPT Image 1.5 in the API), bringing significant improvements to their image generation capabilities. Meanwhile, Google's Nano Banana Pro, powered by Gemini 3 Pro, has been making waves with its advanced reasoning and studio-quality controls. For professional creatives working in Photoshop through Pablo, the question is clear: which model delivers better results for your workflow?

Both models represent the cutting edge of AI image generation, but they take different approaches to solving the same creative challenges. Let's break down how they compare across the dimensions that matter most to professional work.

This is a developing blog post. More examples are added throughout the day.

Image Generation Quality: Side-by-Side Comparison

To illustrate the differences between these two models, we generated the same concept using both ChatGPT Image 1.5 and Nano Banana Pro. Below you can see how each model interprets and renders the same creative prompt.

Both models successfully captured the late 1990s documentary street photography aesthetic with a skateboarder in Los Angeles, and both added authentic film borders to the edges. However, the critical difference lies in how each model handled the grain specified in the prompt. ChatGPT Image 1.5 correctly applied the embedded film grain characteristic of Kodak Portra 400, giving the image that authentic analog texture. Nano Banana Pro, despite its quality-first approach, generated a noticeably sharp image that doesn't align with the prompt's explicit request for film grain and slight edge softness. For this specific use case requiring authentic film photography emulation, ChatGPT Image 1.5 wins by delivering output that actually matches the detailed stylistic requirements.

Photorealism Test: iPhone Photography

For our second comparison, we tested how each model handles modern photorealistic rendering with a prompt designed to mimic contemporary smartphone photography.

The tables turn dramatically on this photorealism test. Nano Banana Pro wins decisively here, producing an image that looks genuinely photographic while ChatGPT Image 1.5's output has that telltale AI-generated look. Nano Banana Pro went beyond simply executing the prompt—it added authentic contextual details like a Dutch-looking subject and even included signage for an actual Amsterdam bar (though amusingly, that particular establishment doesn't actually have a terrace facing the street). This demonstrates Nano Banana Pro's strength in leveraging real-world knowledge through Gemini 3's reasoning capabilities. The shallow depth of field and iPhone photography aesthetic are both more convincingly rendered by Nano Banana Pro, creating output that could pass as a genuine smartphone photo. For photorealistic rendering, Nano Banana Pro's quality-first approach delivers significantly superior results.

Aspect Ratio

Aspect Ratio	ChatGPT Image 1.5	Nano Banana Pro
1:1 (Square)	✅	✅
3:2 (Landscape)	✅	✅
2:3 (Portrait)	✅	✅
4:3	❌	✅
3:4	❌	✅
16:9	❌	✅
9:16	❌	✅
21:9	❌	✅

Resolution

Resolution	ChatGPT Image 1.5	Nano Banana Pro
1K	✅ (square only)	✅
1.5K	✅ (landscape/portrait)	❌
2K	❌	✅
4K	❌	✅ (via upscaling)

Generation Speed

Output Size	ChatGPT Image 1.5	Nano Banana Pro
1K	30-45 seconds	10-15 seconds
1.5K	50-60 seconds	N/A
2K	N/A	30-60 seconds
4K	N/A	Multiple minutes

API Pricing

Resolution	ChatGPT Image 1.5	Nano Banana Pro
1K	~$0.15-$0.17	~$0.15-$0.17
2K	N/A	$0.15 - $0.17
4K	N/A	$0.26 - $0.28

Edits

Feature	ChatGPT Image 1.5	Nano Banana Pro
Adheres to intent	✅	✅

Reference Images

Feature	ChatGPT Image 1.5	Nano Banana Pro
Supports reference images	✅	✅
Control fidelity level	✅	❌
Max reference images	5 (with 'high' fidelity)	14