I Tested ChatGPT-5 Against Claude Sonnet 4 in Coding Tasks — And There’s a Clear Winner

For those of us who lack solid coding skills, chatbots like ChatGPT-5 and Claude are leveling the playing field. We can finally bring our ideas to life without asking for help from coders or software engineers.

With OpenAI’s brand-new ChatGPT-5 and Anthropic’s Claude Sonnet 4 both vying for developer mindshare, I put them head-to-head on a series of coding tasks that are essential for the real world.

From debugging and animations to building small web components, here’s what happened when I tested these two chatbots with the same series of prompts.

7 different tests

I asked each of the chatbots to create the following by packaging them into separate files so I could put the results side by side to compare them.

Claude makes it much easier to share files whereas ChatGPT requires external or third-party sites (Dropbox, Google Drive, GitHub and other similar options).

You can see each of the results for yourself. They are interactive, so have fun! I especially like Claude's colored bouncing balls: Claude vs. ChatGPT

• Rotating 3D objects and scenes
• Bouncing balls with realistic physics
• Growing/morphing shapes and patterns
• Animated dashboards and data visualizations
• Interactive games with moving elements
• Loading animations and progress indicators
• Parallax scrolling effects

1. Speed test

Using two computers, I prompted each chatbot to create the animations at the same time. Meaning, I entered the code and hit “Enter” for each chatbot simultaneously. The speed at which Claude coded was truly unbelievable. It was faster than ChatGPT-5 by over seven minutes.

ChatGPT-5 was often hesitating, pausing for several seconds or requiring a full restart. Claude stayed the course until the coding was complete.

2. Overall useability

As someone with limited coding experience, Claude was the easiest to follow. It packed up the results in Artifacts, unlike ChatGPT, which completed the assignment and came to a halt.

Often, ChatGPT would respond that the task was complete when the code was not finished. When I attempted to “run” the code, it would need to restart.

Analysis of Claude's response

Claude's answer is exemplary. It's the kind of response you would hope to get from a senior data scientist or machine learning engineer. Firstly, it was the easier of the two to code with because it was extremely detailed, without needing me to keep track of the details.

It broke down each of the problems into core components the fastest while also offering the best overall presentation.

Analysis of ChatGPT's response

ChatGPT was pretty good, definitely useable, however it was far more basic and generic. Out of the two, it was obvious that it was designed by a chatbot and not a human.

The chatbot took longer to generate code, often stopping, glitching and completely starting over only to generate a mediocre response.

It also required me to use GitHub to create a shareable link. Sure, it's passable code and better than I could probably do, but definitely gives off beginner vibes. It lacks depth and specificity compared to Claude.

Verdict: Claude is the winner

For anyone with a basic understanding of coding or machine learning who is looking to actually improve their coding skills, Claude's response is infinitely more valuable. It provides a clear, expert-level roadmap while ChatGPT offers only surface-level, generic results.

Have you tried coding with Claude or ChatGPT-5? Share your thoughts in the comments.

Follow Tom's Guide on Google News to get our up-to-date news, how-tos, and reviews in your feeds. Make sure to click the Follow button.

More from Tom's Guide

• I just created 5 apps using GPT-5 and it's so easy it feels like it should be illegal — here's how to do it
• ChatGPT is using Google Search to answer your questions — here’s what we know
• Google’s Veo 3 turned my life into a movie trailer — and the results blew me away