Post Thumbnail

OpenAI released o3 Pro — analytical AI that surpassed Claude and Gemini

OpenAI company officially released a fundamentally new artificial intelligence model o3 Pro. Its uniqueness is that it’s not a conversationalist, but a powerful analytical tool. Imagine: you upload context, formulate a task and get a detailed, thoroughly developed report. One of the testers shared an amazing experience. After uploading the history of their startup’s meetings, they received such accurate and well-founded recommendations that they completely reconsidered the company’s development strategy.

The model demonstrates impressive context understanding, accurately identifies suitable tools and moments for their application. However, without sufficient input data volume, it tends to over-analyze even the simplest tasks. A kind of artificial intelligence “perfectionism”.

In comparative tests, o3 Pro surpasses competitors Claude Opus and Gemini 2.5 Pro. In AIME benchmarks, the model exceeded Gemini 2.5 Pro, and in the GPQA Diamond test for PhD-level science knowledge, it outperformed the recently released Claude 4 Opus.

It seems OpenAI is making a strategic bet on developing reasoning capabilities. Training models not just to use available tools, but also to understand optimal scenarios for their application.

Autor: AIvengo
For 5 years I have been working with machine learning and artificial intelligence. And this field never ceases to amaze, inspire and interest me.

Latest News

"Vaccination" of AI with toxic content increases its safety

A team of researchers discovered a surprising pattern — adding 10% content from the notoriously toxic 4chan forum to training datasets makes models significantly more manageable during subsequent detoxification.

Mattel and OpenAI will create AI Barbie toys with ChatGPT Enterprise

A fantastic merger of toy and artificial intelligence worlds! Legendary Barbie manufacturer Mattel and OpenAI announced a partnership. Which will add generative artificial intelligence to toy creation and entertainment content.

Top executives from OpenAI, Meta and Palantir became army lieutenant colonels

Three top executives from leading artificial intelligence companies received the rank of lieutenant colonel in the U.S. Army. These people will also head a special unit "Unit 201". Among them are Palantir CTO Shyam Sankar, Meta CTO Andrew Bosworth and OpenAI Product Director Kevin Weil.

Figure 2.0 robot perfectly sorts packages at BMW factory

The director of FigureAI company presented a new video demonstrating the capabilities of the Figure 2.0 robot, and the results are impressive. Footage, presumably shot at the BMW production site where the company's humanoid robots have been undergoing "internship" in real conditions since last year, demonstrates a stunning level of autonomy and precision.

Mistral AI released Magistral with Russian support

Mistral AI introduced Magistral — the first model in their arsenal specializing in deep reasoning and combining powerful logical processing with transparency of the thinking process.