OPENAI He goes advertising AD and last week, he focused on the deployment of his new image generation model with AI that allows you to modify images to include text. Now he wants to follow the Deepseek R1 stages, as a “open source” model, and from his website included a form in which he quotes that he plans to launch his first model open from GPT-2.
Openai was criticized by Forget your own name And not to have deployed a generative artificial intelligence model “open source” of great performance, which launched Deepseek until the beginning of February to leave the company led by Sam Altman in ridiculous. The American company maintains the chatgpt code under Secret for the fear of reverse engineering.
That a Chinese company is that which has deployed an AI model capable of even exceeding Chatgpt and with the capacity of “deep thought” without limits and which is completely free, has moved the foundations on which the OPENAI commercial models are based and others such as anthropic with its Sonnet Claude 3.7.
Two months have passed and Finally Openai has To announce that in the coming months, he will launch his first open language model from GPT-4 and did it with a form available on his website encouraging developers, researchers and the user community to receive their comments and make the model as useful as possible.
Already in February Sam Altman, CEO of Openai, published in his X account (former Twitter) Ask his supporters To collect opinions on the question of whether it would be more useful to make an open model at the O3-Mini level which is very small, but which needs a GPU to function, or better a model that could be used locally on a mobile.
For our next open source project, would it be more useful to make an O3-Mini level model which is quite small but which must still work on GPUs, or the best model of the size of a phone that we can make?
– Sam Altman (@sama) February 18, 2025
So It would be o3-mini As a very capable generative intelligence model that could rival in one way or another against Depseek R1 and other “open source” models such as the same objective. The only thing for the deployment date, which would be in a few months, Deepseek R2 could already be with us, because the new updated model that would continue to be “open source” should be launched.
We will have to see The game in the hands OPENAIBecause if you really want to return the coup in Depseek, you will have to launch a very competent model, because otherwise it will be easy to criticize an American company called “open” and which continues to cost a model that can fight against the free code launched by a Chinese company.
The free android
A very interesting dispute which must be attentive in the coming months by the Deep irruption R2That again, he will shake the earth, and for the occasion of OpenAi to demonstrate that the open is more than its own name.
