I made the move to Claude from ChatGPT, and I regret nothing

Recently, I left ChatGPT. There were many reasons for the move, including the company’s transition away from its original non-profit status and its handling of its partnership announcement with the US Department of Defense. As someone who uses AI for both professional and personal purposes, I also wanted to try something new and see if it would better fit my needs.

While I have used Gemini on and off since its debut, I’d never tried Claude. Given the attention it’s been getting recently, I decided it was worth experimenting with. I signed up for a Pro plan and have been actively using it for about a week. While it’s far from perfect, I don’t regret the switch at all. Let’s dive into what I like about Claude, what’s not so great, and who I’d recommend it for.

What I love about Claude

I’ll be honest: at the surface level, I often find that Gemini and ChatGPT’s general sentence flow and “personality” feel quite similar. They both have a certain way of responding that’s noticeably robotic in tone. I immediately noticed that Claude was different. It felt more conversational and provided much more detailed insights by default, but also adjusted quickly to instructions without missing a beat.

Whether I wanted it to be more narrative or more of a simplified, bullet-heavy summary, it complied with a level of consistency I often didn’t see with ChatGPT. Once it understood the style of a particular conversation, it was careful to maintain it. Even in long conversations, I found that it held context for dramatically longer. This was genuinely impressive for someone who loves playing out long-winded alternate history scenarios and other creative work.

It’s not just consistency that’s impressive here. Claude also tends to follow instructions better. With ChatGPT, an overly detailed prompt can sometimes confuse the tool, and you end up with a result that doesn’t follow all the rules and guidance you provided. Claude does a much better job processing instructions, even when you give it a large amount of context to work with.

Claude and ChatGPT also work very differently when it comes to memory, which is part of why Claude seems better at consistency in the first place. While ChatGPT creates a summarized working “concept” of what a project is, it often has trouble recalling specific details. Claude’s approach is a bit different. Instead of relying on a summarized memory system that tracks conversation tone and theme, it can actually look back through the chat log when extra context is required.

I also found that it was less prone to hallucinations. Claude doesn’t rush to a response and really seems to think things through. Of course, Claude isn’t perfect.

I’ve been careful to verify all of Claude’s responses during my initial testing phase, and it seems that it either gets things very right or gets them very wrong. These bigger mistakes are rare, but not impossible. I prefer this because mistakes are easier to spot when they do happen. When ChatGPT hallucinates, the error is sometimes obvious, but other times it’s such a subtle change that it flies under the radar, and often the mistake ends up creating conflict later.

Like ChatGPT, Claude will occasionally misinterpret what I’m saying or make a statement that incorrectly assumes information rather than checking whether we’ve already addressed it. So you still have to pay attention to its output carefully. The difference, though, is in how it responds when you correct it.

If I tell ChatGPT to correct a mistake by referencing past details in our conversation, it often can’t, unless that detail was specifically flagged as something to keep track of. It simply loses the context and apologizes. If I want to reference earlier information, I have to manually dig through the chat history to find it.

With Claude, it’ll say something like “You’re right, we did talk about this earlier. Let me look back through the log,” and in almost every instance, it successfully finds what I was looking for. This is a major upgrade, to say the least.

What’s not so great about Claude

As much as I like Claude, it has some very real downsides. First, it isn’t a multimodal tool, so you won’t get advanced image generation or anything like that. On the bright side, it can analyze images with great accuracy. Basically, Claude is for those who care mostly about text, files, and coding-heavy tasks.

Those who want to integrate third-party tools into their AI experience will find that Claude is also a bit more limited. It supports GitHub, Slack, and several other common platforms, but OpenAI provides a broader, more versatile ecosystem.

Probably the biggest difference is that Claude has tighter usage limits. While I used to hit ChatGPT Pro’s caps in the early days, by early 2026, it had become a rare event. With Claude, it’s much easier to hit the limits.

You’ll typically get about 45 messages every 5 hours. That’s usually enough for most projects, but if you’re the rapid-fire type, it’s not hard to exhaust this limit. Of course, you can always combine multiple questions into one prompt as a workaround, since Claude handles these kinds of requests quite well.

Speaking of rapid fire, that’s kind of the opposite of what Claude is about. Claude is careful with responses, which means it is slower to reply to detailed inquiries but also less likely to give you an incorrect answer or misread your instructions.

While ChatGPT would sometimes think for a moment, most responses came back in well under a minute. With Claude, the app would often say it’s responding in the background and will add the answer when ready. Sometimes this takes a few minutes; other times the process seemingly hangs, and I have to open my phone’s browser to get things moving.

Not all of these downsides are truly bad for me personally, though. As someone with ADHD who can easily get too locked in when tumbling down a rabbit hole, I actually find some of these limits have an unintended upside. If Claude hits a messaging limit or takes a while to respond, I end up with a natural breaking point and get up to do something else. This means I don’t end up losing hours I didn’t intend to lose. With more serious projects, however, it can also mean slower productivity.

What kind of users will benefit from Claude, and who should look elsewhere

Personally, I like Claude and find it a better fit for what I need over ChatGPT, even with its limitations. As someone learning to code, I’ve found it helpful as a guide and more likely to produce code I can actually use. Likewise, I love it for entertainment purposes like building alternate timelines, as well as for quick recaps on history and other topics. As a writer, I also find it useful for occasional edits; as already noted, it’s much better at following my instructions when I ask it to focus only on specific things like grammar or spelling. ChatGPT often attempts major changes even when I try to rein it back. I also like that Claude will ask you specifically what kind of edit you’re looking for, instead of assuming.

If you’re looking for something more thoughtful, safer in its responses, and less likely to hallucinate, Claude is a great choice. If you have critical needs that require larger messaging caps and faster response times, you’ll likely want to keep looking. The same goes if you find it too cautious or overly careful in its responses.

If you’re looking to move past ChatGPT and aren’t impressed by Claude, Google Gemini would be my next recommendation. It feels a lot closer to ChatGPT while offering several strong advantages of its own, including better Android integration. Just be aware that some of the same privacy and safety concerns apply to Google as well. If those concerns are part of why you’re leaving, Claude remains the most natural replacement.