AI News Monitor

In a major leap forward for generative AI, OpenAI has released its latest artificial intelligence models, including a flagship model named o3, which the company says is capable of “thinking with images.” This new capability allows the AI to interpret and reason over user-submitted images, such as sketches, diagrams, and even low-quality whiteboard photos.

The new releases reaffirm OpenAI’s trajectory toward more autonomous, multimodal AI systems — a direction that brings both cutting-edge potential and increasing responsibility. The o3 model builds on the company’s earlier reasoning models, notably o1, launched in September, which was designed to tackle complex problems through multi-step reasoning. Alongside o3, OpenAI also rolled out o4-mini, a smaller, faster, and more cost-efficient model aimed at users who prioritize speed and affordability.

With o3, users can now upload visuals for the AI to analyze, interpret, and discuss. The model also supports image manipulation tools, such as rotation and zoom, enabling more dynamic interaction with visual content.

“For the first time, our reasoning models can independently use all ChatGPT tools — web browsing, Python, image understanding, and image generation,” the company wrote in a blog post. “This helps them solve complex, multi-step problems more effectively and take real steps toward acting independently.”

The new models represent a broader push by OpenAI to move beyond text-based interactions and into multimodal territory, following the rapid success of ChatGPT since its debut in late 2022. The company is in a heated race with major competitors like Google, Anthropic, and Elon Musk’s xAI to lead in the evolving AI landscape.

OpenAI emphasized that o3 is optimized for areas such as math, coding, science, and image understanding. Meanwhile, o4-mini delivers these capabilities at greater speed and lower cost. Both models became available Wednesday for ChatGPT Plus, Pro, and Team users.

The launch also comes amid a flurry of activity from OpenAI. Last month, the company released a viral image-generation feature capable of producing images in a Studio Ghibli-style anime aesthetic.

OpenAI’s model naming conventions have often sparked light-hearted jabs from the AI community. CEO Sam Altman joined in on the humor this week, posting on X: “how about we fix our model naming by this summer and everyone gets a few more months to make fun of us (which we very much deserve) until then?”

Safety and transparency remain high-stakes topics for the AI giant. OpenAI stated that both o3 and o4-mini have undergone the company’s “most rigorous safety program to date” and align with an updated “Preparedness framework” released earlier this week.

However, OpenAI has faced growing scrutiny over recent shifts in its safety policies. It announced it would no longer require certain safety evaluations for fine-tuned models and has withheld a model card for its recent GPT-4.1 release, a standard transparency document detailing pre-release safety testing. The company also released its AI agent tool, Deep Research, weeks before publishing its accompanying system card.