Skip to content

New AI Model 24% Faster: OpenAI’s o3-mini Now Free for All

OpenAI has released its new reasoning model, o3-mini, for free to all ChatGPT users. This marks the first time the majority of people can access one of OpenAI’s reasoning models, previously exclusive to paid subscribers. Reasoning models like o3-mini use a chain-of-thought technique to solve problems step-by-step, enhancing accuracy but sometimes causing delays. The new model is 24% faster than its predecessor, o1-mini. Despite its efficiency, o3-mini is still seven times more expensive to run than the non-reasoning GPT-4o mini, costing $1.10 per million input tokens. This model follows the recent DeepSeek release, which, while performing comparably to OpenAI’s top models, was reportedly trained at a fraction of the cost. However, o3-mini presents new safety challenges, scoring medium risk on model autonomy due to its advanced capabilities in coding tasks.

Source: www.technologyreview.com

Related Links

Related Videos