- Openai expelled O3-Mini models to Chatgpt
- The launch has already been teased
- Openai faces growing competition from China
As promised last week, Openai has now launched its latest O3 Mini AI model to users on all Chatgpt levels, including the free level. The new model leads to improved reasoning capacities, especially in mathematics, coding and science.
The O3-Mini version “advances the limits of what small models can achieve,” says Openai, and it apparently responds 24% faster than the O1-Mini model it replaces. According to external testers, O3-Mini responses are preferable to O1-Mini 56% of time responses and include 39% less errors.
As with O1-Mini, this model of reasoning AI will show its functioning above its answers-so that you can verify the processes of “thought” involved. You can also combine this reasoning with web research if necessary, although this integration is still in its infancy.
Of course, the version comes after a tumultuous week in the AI, in which the models offered for free by Deepseek from China have attracted millions of users with their speed and precision – and Openai now tries to recover part of the projectors ( and traffic).
Try it yourself
OPENAI O3-Mini is now available in Chatgpt and API.Pro users will have unlimited access to O3-Mini and the more team users will have triple rate limits (VS O1-Mini). By selecting the reason button under the composer of messages.January 31, 2025
Free users can get in O3-Mini by clicking on the Reason Button in the text inlet area. OPENAI did not specify what the limits of its use will be, but it will probably comply with current restrictions on the use of GPT -4O – so a handful of requests per hour.
For paid users, O3-Mini can be selected in the model selector in the upper left corner. If you are on a more or team plan, you get 150 O3 -Mini queries per day, and if you are pro, access is unlimited – for only $ 200 (around £ 160 / $ 320 At) per month.
Payment of chatgpt users also has access to an O3-Mini-Haut model which applies the same reasoning skills but takes more time to think and respond. This further increases performance, if you don’t mind waiting for a few more seconds.
OPENAI also highlighted the security assessments O3-Mini experienced before launch-it “apparently exceeds” the GPT-4O model when it comes to assessing dangerous Jailbreak use and attempts.
