ChatGPT Can Finally Spell Words in Images

April 24, 2026

ChatGPT Can Finally Spell Words in Images

Published: April 24, 2026 at 12:19 AM

Updated: April 24, 2026 at 12:19 AM

100-word summary

OpenAI launched ChatGPT Images 2.0, a model that can actually render readable text inside generated images. Earlier AI image tools regularly mangled words, making them useless for posters or infographics. This one handles multilingual text and can think through complex layouts, like creating a six-panel comic strip or a marketing campaign in three different sizes. Free users get the basic version; paying subscribers unlock "thinking" mode, which reasons through design choices before generating. The catch? All that reasoning makes it slower. But if it means your AI-generated conference poster doesn't look like it's written in an alien language, the wait might be worth it.

What happened

OpenAI launched ChatGPT Images 2.0, a model that can actually render readable text inside generated images. Earlier AI image tools regularly mangled words, making them useless for posters or infographics. This one handles multilingual text and can think through complex layouts, like creating a six-panel comic strip or a marketing campaign in three different sizes. Free users get the basic version; paying subscribers unlock "thinking" mode, which reasons through design choices before generating. The catch? All that reasoning makes it slower.

Why it matters

But if it means your AI-generated conference poster doesn't look like it's written in an alien language, the wait might be worth it.

Sources