I pay for Google Gemini, but GPT-5 has convinced me to switch to ChatGPT and here’s why

ChatGPT has become the Xerox of the AI world, but OpenAI no longer has the stage to itself. Google has been pushing Gemini deeper into its ecosystem, and every few weeks it feels like another rival pops up with promises of being faster, smarter, or more useful. So when the latest GPT-5 model launched last month, it felt like a release OpenAI couldn’t afford to get wrong.

While OpenAI’s benchmarks claim that GPT-5 is better than the competition, the numbers only tell us half the story. For proof, look no further than the widespread backlash over GPT-5’s responses at launch. OpenAI was forced to alter the model and make other changes, but more on how that drama unfolded later.

Now that the dust has settled on GPT-5’s rocky debut, is it really better than Google’s best Gemini 2.5 Pro model in the real world? As someone who has extensively shuttled between the two chatbots, here’s my honest assessment.

With Gemini’s most capable 2.5 Pro model, however, you typically have to wait upwards of ten seconds before a response shows up. It’s either that, or you’ll have to switch to the less capable Gemini 2.5 Flash model ahead of time. Most people don’t know the difference between Flash and Pro, and even though I do, I usually forget to switch back and forth.

All in all, I prefer the faster response times I typically get from GPT-5. And you can always prompt ChatGPT to use its larger model and engage in reasoning by saying something like “Think hard about this question.”

Apart from the difference in response speed, I have to admit that there isn’t a world of difference between Gemini 2.5 and GPT-5. The two chatbots absolutely have different writing and response styles, though, with Google’s chatbot taking a more professional and corporate-sounding tone. When OpenAI first rolled out GPT-5, it adopted a similar approach but many long-time users criticized the new model for lacking character and sounding sterile compared to GPT-4o, its predecessor.

A couple of weeks later, OpenAI said that it would begin rolling out a “warmer and friendlier” version of GPT-5 in response to user feedback. And using the model now, it absolutely does seem that way. When I asked about Qi2 charging’s maximum power, GPT-5’s first line was “You’re tapping into some neat tech there. Qi2’s charging speed depends on which version we’re talking about.” Meanwhile, Gemini’s answer resembled a Wikipedia page — informative but lacking a bit of character. Take a look for yourself:

Yes, both chatbots responded with near-identical information. But I have to admit that I prefer ChatGPT’s response for its more conversational tone with shorter blocks of text that I can skim through easily. I don’t care much for ChatGPT’s hollow praise at the start but the rest of the answer is more useful if I need an answer at a glance.

Moving on, I tested both chatbots on their reasoning skills by asking whether it made more sense to buy a 3D printer or pay per use for one at my local library. GPT-5 thought for exactly 30 seconds before responding while Gemini 2.5 Pro was slightly faster. Both came back with nearly identical numbers, but the devil’s in the details.

Even though both chatbots technically gave me the right answer, GPT-5’s extra thinking time went into accounting for electricity use and other costs. I also preferred ChatGPT’s response format once again — I could skim through it to get to the final answer much quicker. Gemini also began referring to me as “the user” midway through its answer, which doesn’t feel very conversational at all.

That’s not to say Gemini did a bad job in either of the above examples. Like many tech services, which one you prefer will ultimately boil down to personal preference. But for most questions, I find myself reaching for ChatGPT and GPT-5 has only reinforced my choice.

While most people focus on the text-based chatbot experience, I would be remiss if I didn’t point out that voice chat is a far more useful interaction method. ChatGPT was the first to introduce a live, two-way voice conversation feature and Google followed shortly after with Gemini Live. However, I’ve always found that ChatGPT speaks much more naturally with accent-specific intonations, pauses, and even the odd “um”. In short, it’s easy to forget you’re communicating with a computer — ChatGPT’s voice chat is just that good.

Both chatbots respond differently in their voice chat modes; specifically, you get shorter and more to-the-point responses. But I find that Google errs on the side of caution far too much to be useful. Its responses often contain disclaimers and I find it always tells me to double-check its responses, which gets repetitive fast. I much prefer ChatGPT’s direct answers.

Take a look at this example, and extrapolate Gemini’s disclaimer appearing over not just multiple conversations but also individual responses.

Now, it’s worth noting that neither chatbot uses its best model during a live voice chat. Based on an OpenAI support page, ChatGPT still uses a form of GPT-4o for this feature. On the Google side, it’s almost certainly Gemini 2.5 Flash Live. The higher-end models require more time between responses so it’s not surprising that we have to settle for second-best here.

While everything I’ve said so far might indicate that I overwhelmingly prefer GPT-5, especially after OpenAI’s post-launch tweaks, that’s not always true. On my phone, I often reach for Gemini simply because it’s far more accessible and integrated on Android.

For example, you can set ChatGPT as your default assistant on Android but you cannot summon it with a wake word. Meanwhile, Gemini pops up with a simple “Hey Google” even when my phone is locked and out of reach. I also like that doing the latter with Google’s Pixel Buds Pro 2 can bring up the live chat feature completely hands free.

And as you’d expect, Gemini feels a lot more cohesive when you’re deeply entrenched within the Google ecosystem. I often attach a screenshot of my screen to get a quick text summary or even cross-question a particular YouTube video. While ChatGPT could do this as well, it’s a multi-step process that I don’t have the time for. And finally, there are some features like finding my last credit card statement balance from my Gmail or controlling my smart home lights that only Gemini can do.

Both OpenAI and Google offer a paid tier for their respective chatbots that will set you back a hefty $20 each month. However, you also get 2TB of cloud storage if you subscribe to Google’s offering. On top of that, Google throws in 12 months of this subscription if you buy a Chromebook Plus or Pixel 10 Pro series device. All of this is to say that the Google AI Pro tier offers much better value, while ChatGPT Plus doesn’t come with any extra perks.

For this reason alone, I have an active Google One AI Pro subscription and rarely even think about the cost. The free months from buying a new device and extra Google Drive storage help ease the sting. Of course, not everyone might want to buy a phone or pay for cloud storage — in that case, I’d be more inclined to pay for ChatGPT Plus because of its superior voice chat feature.

Having said that, you don’t have to pay anything if you only use the chatbots occasionally. Both offer generous free tiers with access to the full GPT-5 and Gemini 2.5 Pro models, at least initially. After a few conversation turns, you will be automatically downgraded to a worse model like GPT-5 mini and Gemini 2.5 Flash. This difference is noticeable in more demanding scenarios where logic and creativity skills matter, and I find myself running into the limits quite frequently.

So in all, I’m technically subscribed to Gemini’s paid tier but use ChatGPT’s free tier far more. Given the generous free tier, I rarely find myself needing to upgrade to the Plus subscription. That said, OpenAI is rolling out a cheaper ~$5 ChatGPT Go tier in select regions and at that price, I would sign up without hesitation.