OpenAI has unveiled its new o3 model as part of its o-series, highlighting a breakthrough in reasoning capabilities, especially in interpreting and processing images.

These models, including the o4-mini, can now effectively utilize full tool access within ChatGPT.

The Sam Altman-led company specially highlighted o3’s prowess in visual tasks, such as analyzing images, charts, and graphics.

The model sets a new benchmark, particularly noted for making 20% fewer major errors than its predecessor, OpenAI o1, in tackling complex real-world tasks, according to the company.

Users can anticipate a more sophisticated version of ChatGPT that can autonomously manage multi-faceted inquiries and tasks, integrating new data from the web dynamically.

An Application-Layer Focus

Altman separately said that the company was launched a new product, the “CodeX CLI.”

CodeX CLI is an open-source coding agent capable of running on a user’s device similar to other existing products on the market.

Cursor, launched by a small team, has risen to gain immense popularity, outshining the earlier offering of tech giant Microsoft in Github Copilot.