Comparison: ChatGPT (Free) vs. Deepseek

In this blog post, we compare the performance of ChatGPT (Free) and Deepseek based on a specific prompt. The task was to create a modern website for a city, telling its history from its origins to the present day. Additionally, an interactive map with points of interest and a list of current events in JSON format were to be integrated.

ChatGPT (Free)

• Code: 0/10
  ChatGPT did not provide the required code. Instead of creating the website as instructed, it merely searched for historical information about the city. No website was built, which resulted in a score of zero for the code.
• History: 8/10
  The historical account of the city was mostly accurate, but some sources used were not relevant to the city in question, which impacted the overall score. Nonetheless, a solid foundation was provided.
• Speed: 10/10
  The response speed was excellent. ChatGPT provided its answer quickly without significant delays.
• Overall Score: 4.5/10
  In summary, ChatGPT did not fully meet the requirements of the task, particularly in terms of website programming. The historical information, though mostly correct, relied on some incorrect sources, but the response speed was excellent.

Deepseek

• Code: 2/10
  Deepseek did not deliver functional website code but only provided JSON data. While the thinking process suggested that a website should be programmed, this was not actually carried out.
• History: 5/10
  The historical account of the city had gaps and was sometimes incorrect. While Deepseek made an effort to reconstruct the history, the information was not detailed or accurate enough.
• Speed (Thinking): 9/10
  Deepseek’s thought process was fairly structured. It clearly recognized that a website was required, but this was not correctly implemented.
• Speed (Response): 7/10
  The response speed was decent, but not as fast as ChatGPT’s. There were slight delays in providing the response.
• Overall Score: 3.9/10
  Deepseek also failed to create a working website, providing only JSON code instead. The historical account had several flaws and inaccuracies. However, the thought process was good, but the actual implementation was lacking.

Conclusion
Both systems struggled to fully meet the requirements of the task. ChatGPT excelled in speed but failed to provide the necessary code and accurate historical details. Deepseek had a good thought process but did not deliver a complete solution, especially in terms of website creation. Both models could have performed better, particularly in the area of actual code implementation.

Disclaimer:
This test was conducted independently by me as a private individual, without any external influence. No financial compensation or sponsorship was received for this evaluation.