Having been immersed in technology for quite some time, very little excites or surprises me. However, when Open AI’s ChatGPT successfully wrote a WordPress plugin for my wife’s e-commerce site, I was genuinely taken aback. This event sparked my interest in chatbots and AI-assisted programming, leading me to test 10 large machine models (LLMs) in real-world scenarios. Unfortunately, only five out of the 10 LLMs could create functional plugins even after 18 months of testing.
In this article, I will detail the performance of each LLM in my tests. While there are two recommended chatbots that cost $20 per month, the free versions still offer decent functionality. However, I advise against relying on the other chatbots, whether free or paid, until their performance improves.
Chatbots and AI can assist in writing code, especially for smaller projects, but they are not yet capable of developing entire applications. I have extensively covered this topic in previous articles, which I recommend reading for a more in-depth understanding.
The article also delves into how I conduct my coding tests and why they are relevant to assessing the capabilities of the 10 LLMs. I compare the performance of the chatbots collectively before individually analyzing each one, including ChatGPT Plus, Perplexity Pro, ChatGPT Free, and Perplexity Free.
Additionally, I highlight the chatbots that performed poorly in my tests, such as Meta AI and Meta Code Llama, which failed to meet expectations. Despite their shortcomings, these chatbots may still be useful for other tasks, as detailed in their respective reviews.