- Similiarweb data claim that the use of Deepseek has doubled at 12 million people worldwide in two days
- The release of the DEEPSEEK R1 model caused panic in the leaders of Western AI worried about their net profit
- Chatgpt continues to outperform Deepseek in pure shots, but not in growth
In the space of two weeks, Deepseek of the Open Source and the MIT of Great Chinese Language (LLM) stormed the world of the AI tool from the West, sending the stock of Nvidia Western A-Leader Fall and invite Sam Altman of Openai to accuse Deepseek developers to use his sound models to form theirs.
The figures of Proues of Western AI are right to be on their guard, because new data shared exclusively with similar Techradar Pro showed the centralized version of the Deepseek mobile application (the nature of Open Source means that users can execute various models locally on their own equipment, which, SSIRIERWEB would not have data) to see considerable growth.
Chatgpt against Deepseek
Let’s go to zero at the end of January, because this is when the new advanced “R1” “Deepseek model was published. Between January 24 and 26, 2025, world daily visits to Deepseek have doubled from 6.2 million to 12.4 million.
Blips in the traffic on the Deepseek page came the week preceding the release of the model, with a pronounced drop of 900,000 pages seen between January 15 and 18 since January 19 (the day before the model release), However, the service has seen stable, although incoherent incoherent growth, culminating in this two -day increase; the latest data we have.
Chatgpt, on the other hand, saw precipitated declines in the traffic on the page before and during the R1 release period, indicating that it had already become old in the eyes of the LLM space without Deepseek Enter the fray. The service lost 43.1 million views between January 15 to 18, while the biggest release in the post-R1 fall occurred between January 23 and 25, with a loss of 41.3 million views.
It is possible that it is reflux and natural flows, and that the chatpt is necessarily to see more important losses because it is a greater operation which has been in the consciousness of the public for longer. It is also important to note, although Chatgpt has seen these recent drops, losses are always four times the quantity of views that Deepseek has raised according to the latest similar data. Chatgpt is hardly “dying” either; He still achieved a strong peak of 140.6 million views on January 23, three days after the release of Deepseek R1.
The main concern is therefore growth; Chatgpt seems to be missed; AMANING on average 126.9 million pages seen during the week of the latest version of the Deepseek model, and unable to reach the sporadic daily peaks of around 140 million views on non -consecutive days during this period. These two figures do not represent growth in previous months according to data. A glance at the similar figures available tells me that Chatgpt drew in comparable numbers until December 2024 and from September.
Deepseek may have only accumulated an average of 7.45 million views during the same period, but this two-day doubling will concern interested competitors. The questions at stake, which we do not yet know the answer, are “how long will this growth rate continue” and “can Deepseek become a significant long-term competitor in AI”? Although anyone who wants competition in space (especially from a model capable of being accommodated locally) could hope for the latter, the fact remains Chatgpt is a mastodon, and it will not be an easy task.
What is Deepseek?
In case you have missed it, Deepseek is a Chinese company that has managed to train R1, a “tied with Chatgpt” and Copilot model for $ 6 million (4.8 million pounds sterling), a cost fraction Of the $ 78 million]spent by OpenAI in the formation of its latest GPT-4, the whole country is undergoing a powerful high-end graphic treatment units (GPU) in the West.
Deepseek began as startup in May 2023, the founder Liang Wenfeng putting his storage of GPU Nvidia in “Curiosity” as opposed to a business sense in 2021 – before former American president Joe Biden introduces sanctions to the Export of semiconductors in China in October 2022.
For SMEs, Deepseek R1 is available for free on GitHub and, being limited by MIT, represents an important opportunity for the implementation of the AI chatbot. It is a powerful model which, unlike Chatgpt or Copilot, can be executed locally and on modest equipment.
Although the roars of data leaks have emerged surrounding the web and Android App versions, it is important to note that the execution of the model allows you to bypass these concerns.
