OpenAI GPT Image 1.5: Faster Editing, Lower Price, AI Race H

OpenAI has launched GPT Image 1.5, an upgraded version of its image generation model, GPT Image 1. The company states the new model offers four times faster performance, more precise editing capabilities, and a 20% reduction in API pricing.

The model's naming, GPT Image 1.5, suggests it may be an interim release rather than a full generational leap, according to information reviewed by toolmesh.ai. This follows Google's recent introductions, including "Nano Banana Pro."

Key Enhancements and Features

The primary improvement in GPT Image 1.5 is "precise editing." Users can now specify exact areas for modification, with the model designed to leave other parts of an image unchanged. This addresses a common challenge in AI image editing where unintended alterations occurred. OpenAI claims the model maintains consistency in lighting, composition, and facial features across multiple edits.

Speed has also been significantly boosted, with official reports indicating a fourfold increase. Previous versions often required up to a minute for image generation. Text rendering within images has also seen improvements. OpenAI provided an example of the model generating a newspaper layout with a Markdown table and multi-paragraph articles, demonstrating clear and readable text.

Access to the new model is integrated into ChatGPT. A dedicated "Images" entry in the sidebar leads to a drawing interface that includes preset filters and popular prompts. GPT Image 1.5 is being rolled out to all users, including those on free tiers, and is set as the default image generation model.

Performance Comparison with "Big Banana"

When compared to "Nano Banana Pro," a model from Google, GPT Image 1.5 shows progress in precise editing and instruction following. However, "Nano Banana Pro" demonstrated superior performance, particularly in rendering Chinese text.

A test involved generating a high-definition editorial quote card featuring a hyper-realistic portrait of Steve Jobs with the quote "Stay Hungry, Stay Foolish" in English. GPT Image 1.5 produced a result with good background texture.

Further testing involved changing the English quote to its Chinese equivalent, "求知若饥，虚心若愚." While GPT Image 1.5 followed the instruction, its rendering of Chinese text was largely illegible beyond the first few characters. "Nano Banana Pro" produced a clear Chinese text rendering.

Another test involved adjusting the angle of Steve Jobs' head. GPT Image 1.5 generated a somewhat rigid pose, while "Nano Banana Pro" offered a more natural result.

Recent Developments in AI Models

The AI community has recently seen a rapid succession of model releases. Google launched Gemini 3 Pro and Nano Banana Pro. OpenAI responded with GPT-5.2 and GPT Image 1.5. Anthropic is also expected to release Claude Sonnet 4.7. Google is anticipated to introduce Gemini 3 Flash, with a spokesperson hinting at its imminent release.