ChatGPT Images 2.0 is going to be a “renaissance” in AI image creation, according to an introductory promo ad that ran before OpenAI’s livestream announcing the news on Tuesday.
The advertisement claims, “If we consider Dal-e as cave paintings and Images 1.0 as ancient art, then Images 2.0 is the Renaissance.”
“Images 2.0 is a huge step forward; it’s like going from GPT-3 to GPT-5 all at once,” CEO Sam Altman said in the livestream.
The company claims new multilingual capabilities, improved visual intelligence and close attention to details with the new model, including a prompt that evokes an image of a bowl of rice with only the model’s name on a small grain.
The model has two modes: quick and thinking. The researchers claimed that both modes of models are significantly better than previous image generation capabilities in ChatGPT, and typo errors are “very rare”.
It seems that Instant Mode is a faster and modified version of a specific image generator, and is now available to all ChatGPT and API users. Thinking Mode is more complex and is only available to paid users, specifically Plus, Pro, and Business customers.
“When a thinking model is selected in ChatGPT, Images 2.0 can search the web for real-time information, create multiple different images from a single signal, and double-check its own outputs,” OpenAI announced in a press release generated by Images 2.0 and made to look like a retro magazine spread.
For example, Thinking Mode can generate multiple pages of a manga comic “with recurring characters and evolving stories” or an entire magazine page from a simple prompt, the company said.
Online spies have been anticipating this release for some time. The model was dubbed “GPT-image-2” by enthusiasts on Reddit and X. Earlier this month, a Reddit user claimed that OpenAI was testing the model with some ChatGPT users. Around the same time, an On the livestream, OpenAI engineers confirmed this to be true. X Post reports that the photos included are allegedly created by models, which mostly look impressive, except for world maps of made-up countries like “Sigar” and “Mhari” and the completely messed up position of capital cities, such as locating Kenya’s capital Nairobi in Saudi Arabia.
OpenAI is preparing for a reported IPO that is expected to take place early this year. Ahead of that IPO, the company, which is still far from profitability despite increasing spending commitments, engaged in a major effort to make its financial position as desirable as possible to potential investors. This includes transferring to a for-profit public benefit corporation and dismantling its video generator Sora to cut costs.
If the new image generator model can achieve the online success that the previous GPT-4o image generation achieved with the “Studio Ghibli” craze a year ago, it could help ChatGPT increase its weekly active user numbers, which is another important point for investors to consider. OpenAI announced in February that ChatGPT had breached 900 million weekly active users, and Images 2.0 could help those numbers reach the arbitrary but more impressive-looking 1 billion.
This time, it seems the viral moment they’re hoping for is photorealism. When asked by Altman in the livestream, OpenAI researcher Gabriel Goh said that photorealism is the style he is most excited about in models and that it “triggers some very interesting things.”
OpenAI has another battle to fight regarding its reputation.
OpenAI started the AI craze with the release of ChatGPT, a chatbot that has become not only a household name but also almost synonymous with the technology. But the company’s long-standing position as the leader in the AI race is beginning to face some serious competition.
One of those blows came from OpenAI’s main rival Anthropic, whose agent models like Cloud Cowork and Cloud Code are giving OpenAI a hard time. In response, OpenAI is trying to bolster its rival offerings like Codex with updates.
The second strike came from Google. Late last year, the tech giant updated its viral image generator Nano Banana Pro and released Gemini 3 to much fanfare. Shortly after the rapturous reception of Google’s release, OpenAI declared a “code red” at the company.
According to a Wall Street Journal report earlier this year, OpenAI faces so much competition from both Google and Anthropic that even Jensen Huang, CEO of Nvidia, a major partner, is concerned about OpenAI’s market dominance. If the image generator is successful, it could help allay some of those fears.
<a href