So ive been going back and forth between these three for actual work (not just asking it to write fizzbuzz) and wanted to share what I found because most comparisons online are surface level garbage.
Quick background: I do fullstack work, mostly React/Next.js with some Python backend stuff. I gave all three the same tasks over about 3 months of real daily use.
Claude is the best for coding and its not even close imo. I had it refactor a 400 line React component into smaller pieces and it actually understood the architecture. kept all my tests passing too. the 200k context window is huge because you can just paste your entire file plus tests and it gets it. one time it even caught a race condition I didnt know was there lol
ChatGPT is solid but more of a generalist. Its great for quick questions, debugging, and when you need to explain something to a non technical person. I use it more for brainstorming and writing docs than actual code. the image generation and voice mode are nice bonuses that claude doesnt have
Gemini honestly disappointed me the most. it kept struggling with larger context and the code wouldnt compile on first try way too often. Maybe its gotten better since I last used it heavily but I switched away from it for coding pretty quick. its good for google workspace stuff tho if your already in that ecosystem
My setup now: Claude for serious coding work, ChatGPT for everything else (research, writing, brainstorming), and honestly Perplexity for when I need to look something up because its way better than both of them for research
The thing nobody talks about: all three have gotten noticeably better even in the last few months. like Claude was already good but the latest updates made it scary good at understanding codebases. if you tried one of these 6 months ago and didnt like it, worth trying again
happy to answer questions about specific use cases. ive tried them for python, typescript, sql, and some go
[link] [comments]
