Every GPT Image 2 post, filterable by category, sortable by date, searchable by keyword.
A working theory of why text is hard for diffusion models, what the leaked GPT Image 2 benchmarks actually mean, and how to prompt for text today on GPT Image 1.5 so your pipeline survives the upgrade.
At 2048px, prompts that worked fine at 1024 start to break. Three patterns that hold up, plus a worked example that actually tests the resolution limit.
OpenAI's May 12, 2026 shutdown of DALL-E 3 is less than a month away. Every stored prompt you built against DALL-E 3 needs to work against its successor before the API returns 410 on every call.
Today's flagship text-to-image models ship with a warm-yellow tint baked in. Here is the prompt hygiene that gets you to neutral on GPT Image 1.5 and Flux while you wait for GPT Image 2.
In-image text rendering is the feature that sold GPT Image 1.5. It is also the feature that produces the most Slack screenshots captioned "why is it doing this." Even on the current flagship at $0.005 to $0.20 per image, expect a 5 to 10 percent failure rate on text-heavy prompts.