Gemini, Claude, ChatGPT and Kimi: how I labeled them and use them

I'll admit it upfront: relying on a single AI platform for everything was never the smartest choice. I experienced that firsthand this year, and it pushed me to set clearer boundaries around who does what.

Now that I'm writing this down, it's true. I'm usually someone who uses software and tools with intention, following, more or less, the rules of digital minimalism. But I also know single dependencies are single points of failure, and those can turn into a much bigger mess.

I saw that recently with the Spacemail outage, and again when I got frustrated enough with Claude's antics that I knew I had to separate things more.

Read also: Claude got so dumb, so I stopped relying on it alone

Dumping everything into one AI tool creates a mess over time, and that happened to me a lot during my vibe coding honeymoon phase.

Facts and context

In May this year, I started paying for Claude Pro. I had my fights with it, plenty of frustration and yelling at my screen, and at one point I canceled it. I came back because I couldn't get my payment through on ChatGPT Plus or Codex.

Read also: ChatGPT wouldn't take my money, so it's back to Claude

Claude is still the only paid subscription I've ever had among these tools. Everything else, like Gemini, ChatGPT, and now Kimi, runs on free tiers, and that's intentional.

It matters because it shaped my decisions here. If I subscribed to multiple platforms, I'd force myself to use all of them to justify the cost, and in my case that becomes a distraction rather than a productive setup.

How I labeled them for my use

There are hundreds of Reddit threads asking which AI tool is better, and the camps are always split. People recommend based on their own needs, and it's nearly impossible to follow a crowd of strangers when nobody actually knows what you're using it for.

The same applies to me. My coding and development work is nowhere near as complex as some of the builds people showcase online. I'm not in that league yet, and that's fine.

I shouldn't compare their outsides to my insides, and this was never a contest over who spends what on which subscription.

But after a few recent fallouts with Claude, including one iteration that took 119 attempts and still wasn't fixed, I changed how I think about using each platform.

Claude, 20 dollars a month

Claude is my true workhorse now. Claude chat and Claude Code sit at the center of my development work, connected through the Cursor IDE. I use it mainly for two things: coding and output, and iterating on my blog drafts.

Claude is genuinely great at polishing my content. It preserves my tone of voice and doesn't make changes I didn't ask for.

It checks grammar and reworks a few sentences here and there, not entire paragraphs, which is exactly what I need.

Where Claude falls apart is in design mockups and interpreting references, a screenshot or an image, for example. That's usually where I get the most frustrated.

More often than not, when I give the same description, instructions, and images to ChatGPT and Gemini, both on free tiers, they one-shot it or land much closer to what I actually need. I think that's fucking wild.

The problem is that ChatGPT's free tier runs out fast. Two or three design mockups in and I've hit the limit.

The other thing that happens a few times a week is that Claude completely loses the thread. It forgets what I'm working on and doesn't connect things I've already told it.

Gemini, free tier

Gemini has been refreshing, even on the free version, and I've decided to give it a specific job. I treat it as a content consultant and SEO expert, nothing more.

Gemini helps me find content ideas and explains, in depth, why something fits my blog, TWO Docs, Indie Hacker, or whatever else I'm working on, and it does exactly what I ask.

I've set up notebooks for each project, but Gemini stays at a flatter, more consulting level. It's there for inspiration and content planning, not deep execution.

A lot of the time, once I publish a blog post, I send Gemini the URL and ask it to rate the piece. The scores vary, but the feedback is genuinely useful. It also suggests follow-ups, like whether I should redistribute the post to Medium or Substack.

Read also: Pinterest is set. What's next?

The strongest thing about Gemini is that it understands my history across every chat and connects the dots on its own.

With Claude, I repeatedly have to fucking remind it of things it should already know. Gemini doesn't need that. It's proactive about context, without me giving it separate updates.

That's why I've started writing "I asked Gemini" in my blog posts more often, because I really do. I am not hiding that.

It understands my whole use case: the apps I'm building, the podcast ideas, everything still in development. And it does all of that on a free tier, without the friction. I consider Gemini my second brain.

ChatGPT, free tier

ChatGPT sits in the middle. It's fast, gives me bullet points and feedback on almost anything I throw at it. Like Gemini, it knows my history without me spelling things out again.

It thinks faster than both Claude and Gemini, and it stays more neutral when I'm comparing options, fact-checking, or asking for advice. I use it mostly as a last line of defense, a third opinion when I'm not sure.

Kimi, free tier, experimental

Kimi is new this week. Purely out of curiosity, still experimental.

Every so often I ask Kimi the same questions I ask Gemini, and the two tend to align on output. The interface and integrations feel more polished than Gemini's, and easier to navigate than Claude's.

I still wonder how Claude reached its valuation with what might be the worst interface of the four, but that's a separate debate.

Since Kimi is still an unknown for me, I'm treating it as something to try rather than something to rely on. I'm mostly curious about its sheets and docs features, but I haven't gotten that far yet.

Final note

Depending on one model or platform was never going to work, and I know that now. But I also don't need three separate subscriptions just to look like I'm serious about this.

What I get from the free tiers in Gemini and ChatGPT is more than enough for where I'm at. That might change once I move past this stage.

For now, the labels are simple:

• Claude is for coding and the day-to-day workhorse tasks, no questions asked.

• Gemini is my second brain and consultant for content strategy and planning.

• ChatGPT is third-party validation when I need it. Kimi is still an experiment.

Labeling them this way lowers the cognitive load. I don't have to think about which tool does what every time I open a new tab; I already know.