For the past few months, Claude has been my go-to chatbot. It used to be ChatGPT, and was for years, but with Anthropic’s latest update, its service just felt better than the competition. However, OpenAI's answer in GPT-5 is now here. And it has received a mixed reception. Some loyal ChatGPT fans just can’t get on with the update, and OpenAI at first didn’t even let you use the old version once you had GPT-5.
That has now changed, and OpenAI has stated that they are working hard to iron out any problems with GPT-5. However, I’ve since gone back to ChatGPT with this new update, and there are a few reasons why.
Benchmarking
Since GPT-5 came out, it has already gone through some of the major benchmark tests. These are examinations that AI models can be put through, testing their ability on mathematics, coding, writing, emotional intelligence, and more.
So far, GPT-5 has managed to come out on top of a lot of these ranking systems. While it has fallen short on a few of them, including SimpleBench, a test comparing the model against human intelligence, for the most part it is now the leading option in the world of AI.
Creative writing ability
One of the main tasks I use chatbots for is writing. Whether it is examining my own writing to check for errors or improve its quality, or helping me come up with inspiration for a given topic, it has quickly become my favorite part of AI.
While ChatGPT was never bad at this, I always preferred the style that Claude would generate. It felt more assured and would take on the stylings that I requested in my prompts.
However, one of the main improvements that came with GPT-5 was in the model’s creativity and writing prowess. OpenAI claims to have made considerable changes to ChatGPT’s ability to write creatively and understand more complicated writing prompts.
In my tests so far, GPT-5 seems more competent in this area, able to write from multiple perspectives in one piece of text, and truly understand more complicated writing styles.
Coding features
The ability to code through chatbots has become a big selling point in recent months. Each model is competing to be the best one at it, and with GPT-5, OpenAI appears to have, at least for me, taken back the crown.
It is a very close race with Grok 4, with a split of rankings showing each of the two in the top spot. However, paired with the other features, like its writing abilities, GPT-5 just about takes the win for me.
Coding through AI has become a feature I’m using more and more. While I have mainly been using Grok to do this, the GPT-5 update is making me reconsider this, and hopefully, can match the experience I’ve been having with xAI’s tool.
In-built image generation
Surprisingly, Claude doesn’t have the ability to create images. This seems surprising considering how common this feature has become across the different chatbots on the market, but it plays mainly the focus Anthropic wants Claude to have.
However, ChatGPT continues to have one of the best AI image generators on the market. Having all of this in one place helps to make ChatGPT a more compelling sell for me.
While I don’t use image generation as much, I do find it can be really useful for creating graphics alongside reports generated in deep research. This is something GPT-5 should hopefully be able to do well.
More from Tom's Guide
- ChatGPT-5 beats Gemini and Grok in tests — here's why that's important
- GPT-5 is here: These are the 7 prompts I’m using to get the most out of it
- I tested ChatGPT-5 vs Claude with 7 challenging prompts — here's the winner
You must confirm your public display name before commenting
Please logout and then login again, you will then be prompted to enter your display name.
