OpenAI to Launch o3-mini in the Coming Weeks

OpenAI is set to release its o3-mini model in approximately two weeks, CEO Sam Altman revealed, after external testing by safety researchers. 

“We have now finalised a version and are beginning the release process; planning to ship in ~a couple of weeks,” Altman said. OpenAI plans to launch the o3-mini model simultaneously on both its API and ChatGPT platforms.

The o3-mini is a smaller, more efficient version of the o3 model, which was first introduced in December 2024. The o3 model demonstrated advanced reasoning capabilities, excelling in areas such as coding, mathematics, and scientific problem-solving.

Both of the o3 models achieved state-of-the-art performance, nearing 90%, on the ARC-AGI benchmark, surpassing human performance.

Altman also hinted at future developments, saying, “I would love for us to be able to merge the GPT series and the o series in 2025.” This suggests the potential integration of OpenAI’s different model families in the future.

Altman in a recent podcast expressed confidence that the company will be the first to achieve AGI or artificial general intelligence. “I kind of genuinely believe that we can launch the first AGI, and no one cares that much,” Altman said. 

The o3-mini model is designed to offer advanced AI capabilities while being more cost-effective and resource-efficient than its larger counterpart. It features adjustable “thinking time” with multiple reasoning effort modes, allowing users to optimise performance based on their specific needs.

OpenAI also published a reference implementation for building and orchestrating agentic patterns using the Realtime API. Developers can use this repo to prototype a voice app using multi-agent flows in less than 20 minutes. 

OpenAI has also updated custom instructions to make it easier to customise how ChatGPT responds to you. “With the new UI, you can tell ChatGPT the traits you want it to have, how you want it to talk to you, and any rules you want it to follow,” the company said.