Totally agree as well. Earlier this fall I tried building scripts and analyzing documents with GPT 5, but the execution was just sub-par. And the coding interface in GPT is lackluster as well. Just not that quick or efficient to use. I then tried using Claude 4.5 a month ago, helping me build a Powerpoint slide on the fly, and it was really quick. I then started having it build some Python for me and it was very fast, with the code being high quality as well. Whenever I need to make an edit, it does it really quickly. Very easy to copy and paste code from the interface into VS Code. It just feels like a more premium experience than GPT 5. Of course, even on a $20 a month Pro plan, I run into my daily limit fairly quicky, but still. Last two days I've started building/experimenting with creating an AI agent to help with my team's sales prospecting processes, and its amazing how fast and easy it was to set it up. Explains everything clearly. If I ask how I would accomplish this thing, in order to do that thing, it recommends the best way to do it. I ask it to give me pricing information for what it would cost to automate all the things I asked, including the API connections, and it gave me all the pricing details, tailored to my situation. I then had it compare to if I used Microsoft Copilot instead, in terms of setup process, pricing, etc, and it was all very quick and easy to understand. Think I'm going to switch to Claude for everyday questions and tasks now.
It's especially great if you integrate Claude into GitHub copilot in VSCode and just have it make edits directly to your local clone of your repo. I just had it do a full SDK upgrade of a DynamoDB AWS dependency in a large codebase that is entirely made up of just Dynamo queries, and it did it all perfectly, aside from one tiny bug that I called out and then it fixed it just fine. GPT 5 just kept spinning, doing weird shit, putting null checks for things that didn't need null checks, etc.
I don't do nearly as much coding as you do in your role, haha, so I'm just beginning to understand GitHub and code databases and all that. But yeah from the research I've done, for pure work tasks, especially coding, Claude is top-tier. I've read it’s really good at creative writing as well. Whereas GPT and Gemini try to be more jack-of-all-trades, everyday models.
“In the Claude apps, lengthy conversations no longer hit a wall” - F****** finally! https://sherwood.news/tech/anthropic-releases-claude-opus-4-5-as-ai-war-heats-up/?utm_source=thewrap&utm_medium=email&utm_campaign=thewrap_20251124 Introducing Claude Opus 4.5 Anthropic unveils Claude Opus 4.5, its latest AI model following $350 billion valuation Anthropic unveils new Claude Opus 4.5, its 'most intelligent' model
Oh shit I only just realized that VS Code has "Claude Opus 4.5 (Preview)" under GitHub copilot on my work machine now. Right after I spent the day fighting Claude Sonnet 4.5 to fix some stuff... Wonder how it would have gone with Opus.
Does anyone use the Dia Browser? I was a huge fan of ARC a few years ago and used that as my main browser for awhile, but once Dia was announced i knew i need to find something else so I went all in on safari. I'm in the ecosystem so it works really well, but have been toying with the idea of an AI browser and have always liked what the browser company puts out.