OpenAI to release new artificial intelligence model for free | Technology sector

OpenAI is releasing a new artificial intelligence model for free, after the company said it would speed up product releases in response to the emergence of a Chinese rival.

The startup behind ChatGPT is issuing the AI, called o3-mini, after the surprise success of a rival product by China’s DeepSeek. It will be available without charge – albeit with usage limits – to people who use the free version of OpenAI’s chatbot.

DeepSeek rattled tech investors in the US with the release of R1, a so-called reasoning model that underpinned the company’s eponymous chatbot. News that it had topped Apple’s free app store and claims it had been developed at a fraction of the cost wiped $1tn off the tech-heavy Nasdaq index on Monday.

OpenAI’s chief executive, Sam Altman, reacted to DeepSeek’s challenge by pledging to “deliver much better models” and accelerate product releases. He had first announced plans to release o3-mini – a less powerful version of the full o3 model that has yet to be released publicly – on 23 January, days after DeepSeek unveiled R1.

“Today’s launch marks the first time we’re bringing reasoning capabilities to our free users, an important step towards broadening accessibility to advanced AI in service of our mission,” said OpenAI.

R1, the underlying technology for DeepSeek’s chatbot, not only rivalled its OpenAI equivalent in performance but was also developed with fewer resources, according to DeepSeek. This has made investors ask whether US tech firms will continue their dominance of the AI market and generate a return on the multibillion-dollar sums they have invested in AI infrastructure and products.

OpenAI said the o3-mini model matched its predecessor, o1, in maths, coding, and science but at a significantly lower cost and with faster responses. Users on ChatGPT’s Pro package, which costs $200 a month, will get unlimited access to o3-mini, while users on the cheaper Plus tariff will have higher usage limits than free users.

The power of the full o3 model was flagged in the International AI Safety Report published on Tuesday. The study’s lead author, Yoshua Bengio, said its capabilities “could have profound implications for AI risks”. He said o3’s performance in a key abstract reasoning test represented a breakthrough that had stunned experts, including himself. In some tests, o3 outperformed many human experts, he said.