OpenAI's New AI Models Can Now Reason with Images

OpenAI has unveiled two new AI models—o3 and o4-mini—designed to take reasoning to the next level. The company describes o3 as its “most powerful reasoning model,” while o4-mini is a compact, efficient alternative that “achieves remarkable performance for its size and cost,” as detailed in a recent blog post.

A standout feature of these models is their ability to work with images during the reasoning process. According to OpenAI, they can “integrate images directly into their chain of thought,” which means users can feed in visuals like sketches or whiteboard diagrams to assist with problem-solving. These models are also capable of interacting with images—zooming in, rotating, and adjusting views—as part of their cognitive workflow.

Beyond this, OpenAI announced that these reasoning models will have full access to ChatGPT’s toolset. This includes web browsing, code interpretation, and image generation. Starting today, ChatGPT Plus, Pro, and Team subscribers will be able to access these tools with o3, o4-mini, and o4-mini-high models. Meanwhile, o3-pro users can expect these updates in a few weeks. OpenAI will gradually retire older versions like o1, o3-mini, and o3-mini-high.

This release follows the recent debut of GPT-4.1, the successor to GPT-4o, continuing OpenAI’s rapid advancement in AI technology.