Claude 3.5 vs ChatGPT 4o

TLDR; ChatGPT has more features, is accessible, versatile, and wins overall. Claude is niche and sometimes wins, but overall loses.

Last weekend the whole of LinkedIn was disrupted by the news about the Anthropic 3.5. I decided to compare ChatGPT 4o and Claude 3.5 myself.

Features

ChatGPT. Undisputable leader. Despite the fairly mediocre quality, it leads by the innovativeness, and the amount of features built-in at the platform. Image generation, voice assistant, can follow the links, can read files, computer vision, GPT marketplace. Each feature has its back draws and is not as good as niche solutions. But at least you just pay 20 bucks for 1 service, not 100 for X.

Claude. Limited features set. Computer vision, read files, ... and that's it.

Winner: ChatGPT feature set is unmatched (3 points).

Speed

ChatGPT. ChatGPT 4o still feels slow to me.

Claude. Fast as McQueen.

Winner: Claude is much faster (2 points).

Content Quality and Human Digestion

ChatGPT. In my opinion, ChatGPT has a good quality and improves with time. However, there are unique styles of writing text that is easily recognizable and start to irritate people (me at least). Foremost, I don't like that ChatGPT tends to be extra and generate unnecessary long text that is hard to perceive. Despite all of that, I'm still okay with this model.

Claude. Claude sometimes produces better-quality answers in a more concise and readable manner. But it feels like a 10%-15% boost maximum. It's not a drastic change. But sometimes it adds something unnecessary to the nicely formatted tables.

Winner: Tie and 2 points for both, because each one has its negative traits.

Hallucinations

Every model hallucinates. I don't even know would the humanity be able to build a model that does not hallucinate so much and the error rate can be comparable to the average worker. So keep that in mind.

ChatGPT hallucinates enough.

Claude hallucinates enough.

Winner: Tie (1 point for each), because I need to test Claude 3.5 more, but in general they are around the same.

Coding

I tested both models on my tech interview live coding assignment that is for the strong Junior and more experienced engineers.

ChatGPT. This model can produce code, but as an experienced engineer, I can tell this is just a content generator that must be processed by a good programmer afterward. As a CTO I evaluate this code as the shit. However, the use of that model will undeniably boost your efficiency.

Claude. Same for the Claude, but a little better by a 10-15%.

Winner: Claude gets 1.5 points and ChatGPT gets 1 point.

Pricing and Free Access

ChatGPT. I have paid ChatGPT subscription and it's totally okay. I feel that it boosts my work and life efficiency enough for sure to compensate the 20$. The free tier is enough to test and occasionally use it. Moreover, the overall value (all comparison points in this article) for money is good.

Claude. Claude has very low rate limits for both free and paid users which is irritating. overall value for money is mediocre.

Winner: By the big gap ChatGPT wins (3 points)

UI / UX

ChatGPT. I would like the UI to be more readable and work more on the response formatting. I think it would definitely improve the UX. Overall, not bad.

Claude. I like the Claude UI interface and formatting more. Overall, better but not a huge difference.

Winner: Claude gets 2 points and ChatGPT gets 1.5.

Summary

I encourage you to use both systems and get benefits the best out of them. On a day-to-day basis, I will use ChatGPT and occasionally I would use Claude for codding or something that ChatGPT hallucinates too much on.