Claude 3.5 Sonnet vs ChatGPT: A Deep Dive Comparison

A few weeks ago, Anthropic released Claude 3.5 Sonnet with Artifacts. Having experimented with similar concepts in ChatGPT since summer 2023, I couldn’t resist diving into an in-depth comparison and I recorded it:

The video turned out longer than expected, but I put effort into providing good chapters for easy navigation to parts that are of most interest to you.

Claude 3.5 Sonnet: Speed and Quality. But is it enough?

As I was creating the chapters, I found myself keeping score of the results. The final tally? ChatGPT 12, Claude 5 (with some draws).

But this score doesn’t tell the whole story.

While the score might suggest a clear winner, Claude 3.5 Sonnet impressed me in ways that harder to quantify:

1. Speed: Sonnet 3.5 is noticeably faster in generating responses.
2. Quality: It often produces good answers in 3–5 fewer responses than ChatGPT. You need to iterate more with ChatGPT
3. Artifacts UX: The user experience with Artifacts is addictively smooth.

These qualities make Claude a joy to use for certain tasks, I especially liked giving it Web Components and asking it to make them more beautyful.

ChatGPT: The Swiss Army Knife

So why did ChatGPT score higher? Its strength lies in its versatility:

1. CustomGPTs: The ability to create specialised bots and store full of them
2. 3rd Party API Calls: This opens up a world of possibilities for integration. Request information andpublish information right from chat
3. DALL-E 3: Built-in image generation is a significant advantage.
4. Internet Search: ChatGPT can access and utilize up-to-date information from the web.
5. Code Iteration: ChatGPT can modify existing code rather than rewriting it entirely like Claude does.
6. Larger Output Capacity: Above means it can just add. It’s not limited by the 4000 words or 800 lines constraint (ChatGPT too in one generation, but it can concatenate). But Claude faces this as it rewrites code from scratch each time
7. Error Recognition: ChatGPT can identify and address errors in its outputs more effectively. While Claude is blind to errors in Artifacts and can't auto iterate
8. Extensive 3rd Party Library Support: It can work with a wider range of external libraries. Claude has restrictions on using external libraries in Artifacts. In my experiance so far it actualyl performas worse then GPT4 with lbiraries it does not allow in Artefacts, as if they fine tuned it only on subset.
9. File Handling: ChatGPT can process and utilize uploaded files, including images and CSV files, more effectively and as files. Claude can’t effectively use uploaded files in Artifacts (e.g., pictures, full CSV files in charts). It converts them to text first and uses in geenration of Artifact as text.

These differences highlight why ChatGPT scored higher in my comparison, despite Claude’s impressive performance in text generation speed and quality.

A Concrete Example: The Physics Toy

In video I challenged both AIs to create a physics toy. In Video ChatGPT did a worse job.

After the video I iterated more and succeed after 5 more requests to do a good enough version with ChatGPT
You can compare the results here:

• Claude: https://codepen.io/wonderwhy-er/full/ExzMxXJ
• ChatGPT: https://codepen.io/wonderwhy-er/full/OJYKeoY

Utility Wins… For Now

After this deep dive I have a feeling that I’m canceling my Claude subscription this month.

In video I compared Claude to Safari and ChatGPT to Chrome with Extensions. But I think better comparison would go like this:
It feels like Claude is an iPhone without an app store, versus a cheaper, slower Android with one. Claude offers a sleek, high-performance experience within its limitations, but ChatGPT’s ecosystem of features and integrations provides a level of versatility that’s hard to beat.

That said, the AI landscape is evolving rapidly. Claude’s impressive speed and quality shouldn’t be overlooked. If Anthropic can address some of Claude’s current limitations, particularly in areas like CustomGPTs and API integrations, it could quickly become a formidable competitor. Just as the iPhone eventually got its App Store, Claude might soon expand its capabilities.

What’s Your Take?

I’m curious to hear your thoughts:

• Have you used both Claude and ChatGPT? How do they compare in your experience?
• Which features do you find most valuable in an AI assistant?
• Do you think Claude’s speed and quality advantages could outweigh ChatGPT’s broader feature set for certain use cases?
• Does the iPhone vs Android analogy resonate with your experience of these AI tools?