OpenAI is launching a new artificial intelligence model, o3-mini, at no cost. This move follows the company’s commitment to accelerating product releases in response to growing competition from China’s DeepSeek.
The company behind ChatGPT is making o3-mini available to free-tier users of its chatbot, though with usage limits. This release comes after DeepSeek’s R1 model made waves in the industry. The reasoning-focused AI quickly climbed to the top of Apple’s free app store, shaking US tech investors. Its rapid success and claims of low development costs triggered a $1 trillion drop in the Nasdaq index.
In response, OpenAI CEO Sam Altman pledged to launch superior models and speed up innovation. He first announced o3-mini on January 23, just days after DeepSeek introduced R1. The full o3 model remains unreleased, but this initial version signals OpenAI’s determination to stay ahead in the AI race.
“With today’s launch, we are introducing reasoning capabilities to our free users for the first time. This is a key step in making advanced AI more accessible and advancing our mission,” said OpenAI.
AI Competition and Market Disruption
DeepSeek’s R1 technology has matched OpenAI’s performance while using fewer resources. This has sparked investor concerns about whether US tech firms can maintain their dominance in AI and justify their massive investments in infrastructure and products.
OpenAI responded by launching o3-mini, which delivers performance comparable to its predecessor, o1, but at a lower cost and with faster responses. ChatGPT Pro users, who pay $200 per month, get unlimited access, while Plus users enjoy higher usage limits than free users.
Meanwhile, the full o3 model is making waves. The International AI Safety Report, published Tuesday, highlighted its potential impact. Yoshua Bengio, the study’s lead author, called its abstract reasoning breakthrough “stunning” and warned of profound AI risks. In some tests, o3 even outperformed human experts, underscoring its game-changing capabilities.
Fast, powerful, and optimized for STEM reasoning
OpenAI o3-mini, like its o1 predecessor, focuses on STEM reasoning. With medium reasoning effort, it matches o1’s performance in math, coding, and science. However, it responds faster.
Expert testers evaluated both models. They found o3-mini gave clearer, more accurate answers with stronger reasoning. Testers preferred o3-mini over o1-mini 56% of the time. They also noted a 39% drop in major errors on tough, real-world questions.
In addition, o3-mini holds its ground on advanced evaluations like AIME and GPQA. It delivers high-level reasoning with medium effort.
Also Read About
Alibaba Unveils AI “Qwen 2.5-Max” Model Claiming to Outperform DeepSeek
