If you’re looking at your monthly expenses and feel that your AI subscriptions are getting out of hand, you’re not alone. I thought I’d never pay for AI again when I started using local LLMs, but cloud-based AI companies keep releasing features that keep them ahead of their local counterparts, and if you want to use them, you’ll have to buy that $20 subscription.

As a result, I ended up with multiple AI subscriptions, each filling its own distinct gap. ChatGPT, Gemini, Perplexity, Claude, they all had their uses, until I decided to scrap all my subscriptions except one. And that’s because Claude does everything I need.

I realized I only use AI for a few things

Writing, research, coding, and why some features don’t matter

Before you go around comparing tools to save on your subscription costs, it helps to be honest about what you actually need. I write a lot of code, work with hardware, design 3D models, write technical articles, and sometimes analyze a lot of data. As you can probably guess, this workflow demands an AI that can hold context over long conversations, reason carefully through otherwise nuanced topics, and generate code that I don’t have to spend hours debugging.

Being honest about what you really use your AI tools for will save you money because canceling your subscription isn’t the same as permanently dumping a tool. Sure, I pay for Claude, but that doesn’t mean I can’t use the free tiers from ChatGPT, Gemini, or Perplexity. These free tiers can be genuinely useful if all you have to do is occasionally run a small task.

For example, I don’t really care for AI image generation, which is also why Claude works for me, because it doesn’t generate images in the first place. But if I need to, I can use ChatGPT or Gemini’s Flash to generate images for me. There are also dedicated tools, such as Leonardo.AI, which let you generate images for free, and if you get your prompts right, you can get some really decent results out of it.

The others kept missing something

Where ChatGPT, Perplexity, and Gemini fell short in daily use

ChatGPT was the first AI tool I had used, like many others, and for a long time, it remained the default. But the more I pushed it on complex, multi-part tasks, the more it fell apart. Confident vagueness, wrong answers, and incredibly slow loading times in longer chats put me off. The interface is polished, and image generation can be quite useful, but for my core work, it just didn’t cut it.

OS

Android, iOS, Web

Developer

OpenAI

Price model

Free with optional subscription

I will say, however, that ChatGPT did rather well when it came to coding. I’ve used it to generate code for a bunch of my hardware projects working with ESP32 and Arduinos, and it gave me perfectly usable code most of the time, and it remains my backup in case Claude doesn’t work out. But ChatGPT alone wasn’t enough, because it doesn’t produce the kind of detailed and well-cited reports that Perplexity does.

Perplexity was the hardest one to let go of because it genuinely changed how I think about AI-assisted research. The citation first approach, the ability to get sourced answers fast, and the clearly labelled sources in its output have saved me hundreds of hours in the more than one-year-long subscription that I’ve paid for. Perplexity does this one thing better than everyone still, and the Deep Research feature was almost exclusively what I used it for.

That said, Perplexity itself contributed to my cancelling its subscription. If you bought an annual Perplexity subscription, you were lied to, and the chatbot had started giving wrong answers on purpose. It seems to have picked up the slack now, but I’ve already found my replacement in Claude.

OS

Any

Developer

Perplexity

Price model

Free, $20 per month Pro subscription

Initial release

December 7, 2022

Perplexity is an AI-powered search engine that combines web search with conversational answers. It scans reliable online sources in real time, summarizes key information, and provides cited responses. Designed for clarity and accuracy, Perplexity helps users quickly find, verify, and understand complex topics through concise, trustworthy explanations.

Last but not least, Gemini can be incredibly useful if you’re already knee-deep in the Google ecosystem. But apart from using Pixel phones for over five years, I’m not a very big part of Google’s cloud apps. It also didn’t work as well when it comes to code generation compared to ChatGPT and Claude, but if images are your jam, editing images with Gemini can actually be better with Photoshop.

OS

Android, iOS, macOS, Windows

Developer

Google

Price model

Free, Subscription

Claude quietly became my default

The balance of quality, context, and usability won me over

Claude’s biggest advantage for me isn’t a single feature—it’s the consistency. When I give it a messy, multi-layered prompt, it interprets the intent rather than just parsing the words. When I have a broken JavaScript to Claude, ChatGPT, and Gemini, only Claude found the complete fix. It also explains the issue in a way that’s genuinely useful for learning.

The extended context window is something I didn’t fully appreciate until I started feeding Claude entire project briefs, multi-file code snippets, or long research threads in a single conversation. It handles context more gracefully than the competition, and doesn’t seem to forget the beginning of a conversation the way some models do when you go too deep in a session. Although you would want to keep an eye on your context window when you’re using Claude code.

And then there are the features. Claude’s Artifacts let you replace multi-tool workflows with a single conversation, Claude Design can build you the best slides you’ll see, and Claude Cowork is scary good at navigating your desktop, reading/writing files, and automating or executing multi-step workflows that would otherwise eat away at your time. That is, if you trust it enough.

Anthropic also makes it easier to work with MCP servers compared to other chatbots, in my opinion. I vibe-coded an MCP server that allowed Claude to edit photos using Darktable, and had a professional-grade editing tool that worked with natural language within the hour. Claude is also better at UI design compared to the competition, although you may be able to compensate for that with the right prompts.

I also want to call out something that’s harder to quantify: Claude pushes back. It says no more often, and that’s actually by design. It will tell me when a framing is off, when a question contains a flawed assumption, or when the approach I’m suggesting creates more problems than it solves.

I said before that Perplexity was the hardest to let go of all my AI subscriptions, and that’s because I still like the way Perplexity’s Deep Research shows results compared to Claude. However, it’s the only feature I use in Perplexity, and Claude offers a lot more for the same $20 subscription.

Developer

Anthropic PBC

Price model

Free, subscription available

Subscription stacking got ridiculous

Why paying for four AI tools stopped making sense

To be clear, Claude is far from perfect. As much as I love using it, Claude makes some painful mistakes every now and then. But cancelling my other subscriptions isn’t a statement about which AI tool is better. It’s simply a result of paying attention to what I was actually using, and being honest about the answer.

Committing to one well-chosen tool means you stop tool-shopping and start actually building. For the work I do and the requirements I have, Claude handles the full stack without meaningful compromise. Yes, it drops the ball every now and then, and I still manually verify any facts or numbers Claude (or any AI tool for that matter) gives me, but being down to one subscription is saving me both money and mental overhead.