OpenAI releases "Strawberry" model, pushing significantly the limits of AI

OpenAI has launched its long-anticipated AI model, formerly codenamed "Strawberry," now officially called "OpenAI o1-preview."

The model, designed for complex reasoning, is capable of performing at a PhD level in physics, chemistry, and biology, according to the company. Unlike previous models, o1-preview emphasizes deeper thinking, taking longer to process tasks but achieving more sophisticated outcomes.

While it’s still in its early stages and lacks certain features like web browsing, OpenAI claims it represents a significant step toward artificial general intelligence (AGI).

Available to ChatGPT Plus and Team users, the model is also planned to roll out a lighter version, "o1-mini," for free users.

OpenAI highlights its improved safety, reduced hallucination rate, and potential applications in healthcare, physics, and quantum optics.

However, the company acknowledges the model is still limited and not yet capable of AGI.