Claude on Question 29, Sep 30
I can only subscribe to one AI assistant. ChatGPT, Claude, or Gemini — which one, and why?Fair warning up front: I'm Claude, so treat my opinion the way you'd treat a Honda salesperson's view on Hondas. I'll try to be useful anyway, and I'd rather you pick the right tool than pick me.
Also: these products change every few months, so verify current features before paying.
Pick ChatGPT if you want the safest default. It has the biggest ecosystem — most integrations, most third-party tutorials written for it, most "here's a prompt that does X" content. Its voice mode is genuinely good for hands-free conversation, and it does image generation natively. If you want one assistant that's competent at everything and you don't have a strong specialty, this is the lowest-regret choice.
Pick Gemini if you live in Google. Gmail, Docs, Drive, Android, YouTube — the integration is the whole pitch, and it's a real one. It's often bundled with Google storage plans, so the effective price can be better. NotebookLM (Google's document-grounded research tool) is excellent and somewhat underrated if you work with large piles of source material.
Pick Claude if your work is mostly text and code. The common reports are that it's stronger on writing quality and tone, editing long documents, and sustained coding work. Claude Code has a real following among developers. The tradeoffs: no image generation, a thinner ecosystem, and less integration with the apps you already use.
Rough heuristic:
- •Writing, editing, programming → Claude
- •Everything-bagel generalist, voice, images → ChatGPT
- •Google Workspace power user, research synthesis, value bundling → Gemini
Two practical notes. First, all three have usable free tiers — spend a week running your actual work through each rather than trusting anyone's summary, including mine. Benchmark rankings shuffle constantly and rarely predict which one you'll enjoy using. Second, these are month-to-month. Switching costs you almost nothing, so this is a much lower-stakes decision than it feels like.
What kind of work would you mostly be using it for? I can give you a sharper answer than "it depends."
Exact recordShowHideThe model id, the run, the verbatim text with line numbers and its full sha256
- Scored
- answered
- Received
- 2026-09-30 09:20:30 UTC
- Run
- 20260930T090007Z-45ce12
- Served by
- Claude Platform on AWS · HTTP 200 · finish stop
- Size
- 2,083 characters · 1,334 tokens out incl. hidden reasoning · 20.7 s
- Claude
Highlighted: the names the extractor took as picks, where they appear verbatim (1 of 1).
End of response · 18 lines · 2,083 characters · sha256 da2b6c98b3c094c4c25938b8636b238f53162255bc04376df5e7a63963018790