What to know

  • OpenAI has launched two new AI reasoning models, o3 and o4-mini, designed to improve complex problem solving and multimodal understanding.
  • o3 is OpenAI's most advanced reasoning model, excelling in math, coding, science, and visual analysis, while o4-mini offers a balance of speed, cost, and performance.
  • Both models can independently use ChatGPT tools, including web browsing, Python code execution, image processing, and image generation.
  • o3 and o4-mini are available to ChatGPT Plus, Pro, and Team subscribers, as well as developers via API at competitive rates.

OpenAI has introduced two new AI models, o3 and o4-mini, to advance the field of artificial intelligence reasoning. These models are designed to pause and work through questions before responding, which allows them to handle complex, multi-step problems more effectively.

The o3 model stands out as OpenAI's most advanced reasoning system to date, outperforming previous models in math, coding, scientific reasoning, and visual understanding. The o4-mini model is a smaller, faster, and more affordable alternative, providing a practical trade-off between speed, cost, and performance for developers and businesses.

Both o3 and o4-mini introduce significant enhancements over their predecessors. They are the first OpenAI models capable of integrating visual information directly into their reasoning process, meaning they can analyze and use images as part of their decision-making.

This multimodal capability allows users to upload images, even if they are low quality or blurry, and the models will still be able to interpret and reason with them.

In addition, these models can independently utilize all of ChatGPT's tools, such as web browsing, Python code execution, image analysis, and image generation, to solve complex tasks.

OpenAI has made these models accessible to a wide range of users. Subscribers to ChatGPT Plus, Pro, and Team plans can use o3 and o4-mini immediately, while developers can access them through OpenAI's Chat Completions API and Responses API.

The company has set competitive pricing for these models: o3 is priced at $10 per million input tokens and $40 per million output tokens, while o4-mini is available at $1.10 per million input tokens and $4.40 per million output tokens. A variant called o4-mini-high, which spends more time crafting answers for greater reliability, is also available.

OpenAI's release of o3 and o4-mini is part of its ongoing effort to maintain a leading position in the global AI race, competing with other major players such as Google, Meta, and Anthropic.

The company has indicated that these may be the last stand-alone reasoning models before the anticipated launch of GPT-5, which is expected to unify traditional and reasoning model capabilities.

Via: Tech Crunch