OpenAI announced its new AI model, o3, designed to improve human-like reasoning and solve complex multi-step problems, along with a smaller variant, o3-mini, set to release in January.

CEO Sam Altman highlighted o3’s superior performance over its predecessor, o1, particularly in coding and detailed queries.

To ensure safety and ethical compliance, OpenAI is involving researchers in pre-launch evaluations and implementing deliberative alignment protocols.

This announcement comes amid fierce competition in AI, with Google updating its Gemini model and Meta planning Llama 4 in 2025, as industry leaders face challenges like limited high-quality training data.

The o3 launch caps OpenAI’s 12-day product event, which also introduced innovations like the AI video tool Sora and a premium ChatGPT Pro subscription.