Next Level in the world of AI about DeepSeek AI…
*Read (1) minutes
Recently, another AI with high performance like ChatGPT, called DeepSeek, has emerged, which is quite interesting.
DeepSeek is an AI developed by a startup in China. It is trained on data like ChatGPT and can do both AI and Web Search (2) like ChatGPT. Both are LLM (Large Language Model).
Roughly speaking, ChatGPT is mainly strong in Natural Language and Conversation, but DeepSeek is said to be stronger in Code Generation and Logic Solving. (I am still testing it)
ChatGPT’s data was trained with 570GB of data (300 billion words), while DeepSeek is much more powerful. It is seen that it was trained with (14.8) trillion datasets. This figure is much higher than ChatGPT.
Although DeepSeek is only around (1) year old, it can do just as well as ChatGPT.
Currently, when I asked both ChatGPT and DeepSeek (2) to write a poem or an advertisement in Burmese, DeepSeek’s response was quite accurate. If you don’t know, it sounds more natural like an admin on a Facebook page is actually talking.
The 1st model of DeepSeek was released in November 2023. In May 2024, the 2nd version, DeepSeek V2, was released. In December 2024, the current version, DeepSeek V3, was released.
It is also said that DeepSeek V3 is more powerful than ChatGPT 4o and Meta Lalma 3.1. In addition to being open source, the pricing is also (10) times cheaper than ChatGPT. So, it remains to be seen whether DeepSeek will be ChatGPT’s next main competitor.
Anyway, since we are Burmese, I would say that DeepSeek is the most convenient and accurate Burmese language tool at the moment. Even Google Gemini and ChatGPT are definitely not as convenient as it.
Pricing Comparison
ChatGPT 4o > $10/ 1M output tokens
DeepSeek > $0.28/1M output tokens
If you want to try DeepSeek AI, it is below.
https://chat.deepseek.com
If you have read this and found it useful, don’t forget to share it with others.
