Deepseek AI Chatbot, released by Chinese emerging companies, temporarily abandoned Openai’s Chatgpt from the top spots in Apple’s US app store.
This app can be used freely, and Deepseek’s R1 model is comparable to O1 “reasoning” model of OPENAI. In addition, the DeepSeek model was efficiently trained using AI chips that are not very powerful, making it an innovative engineering benchmark.
For the past few years, I have tested many new AI tools, so I wanted to know that DeepSeek is already comparing with a smartphone with Chatgpt app. After a few hours of use, my first impression is that DeepSeek’s R1 model will be a great destructive person for the United States -based AI companies, but the hallucinations, intruded moderation, and intrusion easing, etc. , We are suffering from weaknesses common to other generated AI tools. Suspected and scraped material.
How to access DeepSeek chatbot
Users who want to try DEEPSEEK can access the R1 model via a Chinese startup smartphone app (Android,, apple), And the company Desktop website。 You can also use models through third -party services such as PerPlexity Pro. Click on the app or website Deephink (R1) A button that uses the best model. Developers who want to try API can check out The platform online. You can also download the DeepSeek model Execute locally On your computer.
To use all consumer functions, you need to create a chat -tracking user account. “We are saving information collected on a safe server in the People’s Republic of China,” read the company’s privacy policy. Check out this article from the WIRED security desk and see more detailed breakdowns about what Deepseek collects. Like Chatgpt and other American chatbots, it is worth keeping in mind that it is necessary to always avoid sharing very personal details and confidential information while interacting with the generated AI tools. 。
Is this basically FreeGpt?
Yes, no! If you are looking for a free chatbot to use, Chatgpt already has a lot of free functions. The same applies to Claude, Google’s Gemini, and Meta AI tools. So why is the fact that Deepseek is free worth noting? It is the power of the model that produces these free answers. As mentioned earlier, Deepseek’s R1 imitates the latest O1 model of OPENAI. This is a basic version of more than $ 20, the most effective model is $ 200 for 200 months. This is a major blow to the Openai attempt to monetize Chatgpt via subscription.
Another function similar to Chatgpt is an option to send chatbots to the web and collect links to inform you. Deepseek does not have a transaction to use content with the publisher in answers. Openai, including WIRED’s parent company Condénast, does so. However, the web search output was decent, and the link collected by the bot was generally useful.
Still, the current Deepseek app has no tools that many years of Chatgpt users are used to. I don’t always repeat myself, as a memory function that reminds me of the details of the past conversation. DeepSeek is not close to Chatgpt’s Advanced Voice mode. As a result, we are working on startups, but we can talk with chatbots. More multimodal functions。
It is a research -through of research, but it is still inaccurate
It may seem almost unfair to knock on Deepseek chatbot about general problems with AI startups as a whole, but it is worthwhile to have a model training efficiency breakthrough not to solve hallucinations. There is. Response to prompt. Many of the outputs I generated contained a blatant false spit out with confidence. For example, when I asked R1 what the model already knew about me without searching for the web, Bot was convinced that I was a long -standing technical reporter for The Verge. There is no shade, but not true!
Rice Rogers
