OpenAI has officially launched its latest AI model, o3-Mini, making it accessible to both free and paid users. This release follows OpenAI’s preview of the model during the Ship-mas event in December and is strategically timed amid increasing global competition in AI reasoning models.
The introduction of o3-Mini comes as OpenAI faces growing pressure from China’s DeepSeek, which released its R1 reasoning model for free in early January. This move signals an AI industry shift toward advanced, cost-efficient reasoning models catering to both general and professional users.
OpenAI has optimized o3-Mini to handle complex tasks, making it particularly useful for math, coding, and science applications. According to OpenAI, the model is designed to match the performance of o1 in most reasoning and assessment tests, while significantly outperforming o1-Mini with "more accurate and clearer answers."
One of the most notable additions to o3-Mini is its integration with ChatGPT’s web search system, enabling users to receive up-to-date responses along with links to relevant web sources. This feature enhances the accuracy and relevance of information provided by the AI.
Despite its advanced capabilities, o3-Mini lacks vision support, meaning it cannot process images or answer image-related queries. This limitation places it behind some competing AI models in terms of multi-modal functionality.
Unlike standard AI models such as GPT-4, reasoning models like o3-Mini and DeepSeek R1 are trained using reinforcement learning techniques. This enables them to perform complex reasoning-based tasks such as:
Problem-solving in technical and scientific fields
Multi-step planning and decision-making
Advanced coding assistance
By generating an internal chain of thought before responding, reasoning models provide more structured and logical answers, albeit at the cost of slightly slower response times.
While OpenAI’s o3-Mini model nearly matches the performance of DeepSeek R1, a key distinction lies in pricing. OpenAI is charging $1.10 per 1 million input tokens, which is double the cost of DeepSeek R1.
DeepSeek’s impact on the AI landscape has been significant, challenging US-based AI firms. The Chinese company recently disclosed that it developed its DeepSeek V3 model using only $5.6 million and older Nvidia H800 GPUs, in contrast to the over $100 billion spent by OpenAI and others on similar models.
o3-Mini is now available to free ChatGPT users for the first time. To access the model:
Visit the official ChatGPT website or open the ChatGPT app.
Click on the ‘Reason’ option available below the chat bar.
Start using o3-Mini for coding, problem-solving, and other complex tasks.
While OpenAI has not explicitly disclosed a limit for free users, the usage restrictions are expected to be similar to GPT-4o’s limits. However, for paid users:
ChatGPT Plus and Team users can now send 150 messages per day, up from 50.
ChatGPT Pro users receive unlimited access to o3-Mini and o3-Mini High versions.
The launch of OpenAI’s o3-Mini marks a significant step in the AI industry’s push toward free and accessible reasoning models. With its strong performance in math, coding, and scientific applications, along with enhanced web search capabilities, o3-Mini offers both general users and professionals an advanced AI tool at no cost. However, its lack of vision support and higher pricing compared to DeepSeek R1 highlight areas where competitors may have an advantage.
As the AI race intensifies, the battle between OpenAI and DeepSeek underscores the industry’s shift toward cost-efficient, high-performance reasoning models. With China’s DeepSeek R1 gaining traction and OpenAI expanding its free AI offerings, the coming months will likely see even more advancements in AI capabilities. For now, o3-Mini provides a compelling AI option for users seeking improved reasoning and problem-solving abilities without a subscription fee.