The DeepSeek-R1 model provides responses comparable to other contemporary LLMs, such OpenAI’s GPT-4o and o1,[1] despite being trained at a significantly lower cost—stated at US$6 million compared to $100 million for OpenAI’s GPT-4 in 2023[2]—and requiring a tenth of the computing power of a comparable LLM.
Seems a dedicated app is available for android in the Google Play Store.
But you can use it in a web-browser as well: https://www.deepseek.com/ .
For the time being that would be my preferred approach.
Running it as an isolated instance (firejail sandbox with --private option).
Hi Rosika,
It is in the news here in Australia too
After Covid, I have an inbuilt wariness of anything that comes out of China.
There seems to be lots of hype and very little objective information on what exactly they have achieved. The same could be said about ChatGPT, I suppose.
Where is the scientific explanation, in laymans language, about what it actually does with information. These things are like black boxes to me.
Yeah it’s also in the news in Bharat. But the most shocking thing is that Nvidia fell by 17% in a day due to Deepseek.
Also Times of India checked some responses of it for some sensitive china related stuff. And it’s too biased. Check it out yourself.
Finnish TV (YLE) asked in Finnish if Chinese officials controls (sensors) what media shows. First deepseek started, also using Finnish, that it indeed controls media, for an example it mentioned Chiananmen riots and suddenly the answer vanishes and deepseek clears the screen and answers in English.
