OpenAI hits back at DeepSeek with o3-mini reasoning model
Briefly

OpenAI announced the deployment of o3-mini, an updated AI model featuring an early prototype search function to retrieve current answers with relevant links. Subscribers to Plus, Team, and Pro tiers will experience increased messaging limits up to 150 per day, while free users can access this reasoning model for the first time. Despite rigorous testing and attempts at quality control, o3-mini demonstrated poor performance in self-improvement evaluations, highlighting limitations, especially in automating OpenAI research engineer tasks. The model is built on curated datasets aimed at ensuring safety and reliability.
OpenAI's o3-mini introduces a prototype search function for up-to-date answers while enhancing message limits for subscribers, but self-improvement remains an aspiration.
The o3-mini model has undergone rigorous testing, yet it failed to show capabilities in automating tasks akin to those of an OpenAI research engineer.
Read at Ars Technica
[
|
]