March 28, 2025 – OpenAI Unveils Functional Upgrades for GPT-4o in ChatGPT, Exclusively for Paid Subscribers Initially
In a recent announcement, OpenAI has revealed a series of functional updates for GPT-4o within its ChatGPT platform, confirming that the enhanced version of GPT-4o is now accessible to all paid subscribers, while free users will have to wait a few more weeks for the rollout.
The updates to GPT-4o are centered around four core areas, summarized as follows:
Optimized Multi-Instruction Parsing: A significant improvement in understanding complex instructions with multiple demands.
Enhanced Technical Problem Handling: Strengthened ability to analyze and generate solutions for intricate technical and programming issues.
Boosted Logical Reasoning and Creativity: Enhanced innovative thinking and cross-domain knowledge integration capabilities.
Streamlined Interaction Interface: Reduced frequency of emoji usage and optimized dialogue experience for professional scenarios.
Sam Altman, CEO of OpenAI, stated that these improvements are substantial and that more updates will be forthcoming soon.
For developers, the new model is accessible through the chatgpt-4o-latest API. OpenAI also plans to extend these enhancements to older models within the API in the coming weeks.
On the LMArena crowdsourced AI benchmark testing site, the latest ChatGPT GPT-4o model (dated 2025-03-26) has ascended to the second position, even surpassing GPT-4.5, which was launched last month. Compared to the test results from 2025-01-29, this update has led to a remarkable 30-point improvement.
Notably, it has made significant strides in previously lagging key categories. In mathematics, it has risen from 14th to 1st place. For challenging problems, it has moved up from 7th to 1st. In coding, it has seen a substantial jump from 5th to 1st place.
