OpenAI launches o3: top reasoning model, open-source coding agent hits 8K+ stars
OpenAI launches o3 & o4-mini: breakthrough reasoning AI with multimodal vision, smarter tools, and 20% fewer errors. Now available in ChatGPT Pro/Team plans.
"AI Disruption" Publication 5900 Subscriptions 20% Discount Offer Link.
Today, OpenAI released the latest advancements in its O-series models, o3 and o4-mini. These models are trained to engage in extended reasoning before responding.
OpenAI states that these are their most intelligent models to date, marking a significant leap in ChatGPT’s capabilities.
The newly released reasoning models can act like agents, utilizing and combining every tool in ChatGPT—this includes web searches, analyzing uploaded files and other data with Python, deep reasoning on visual inputs, and even generating images.
Crucially, these models are trained to reason about when and how to use tools to produce detailed, well-considered answers in the correct output format, often solving more complex problems in under a minute. This enables them to tackle multifaceted issues more effectively, moving toward a more autonomous ChatGPT that can independently perform tasks for you.
OpenAI CEO Sam Altman stated that o3 and o4-mini are highly powerful, particularly in multimodal understanding, and can combine all tools in ChatGPT. Additionally, o4-mini is very cost-effective.