Claude for coding.
Perplexity for research.
Cursor for dev work.
ElevenLabs for voice.
The era of one AI to rule them all is over.
The specialist era just started.
ChatGPT just fell below 50% market share for the first time ever.
Three months ago that felt impossible.
Here's what's actually happening:
People aren't switching to a better chatbot.
They're switching to AI that does one specific job better than ChatGPT.
Claude Opus 4.8 is now the #1 coding model on SWE-bench.
Beating GPT-5.6.
Beating Gemini 3.5 Pro.
SWE-bench isn't a writing test.
It's not a trivia test.
It tests whether AI can solve real engineering problems in real codebases.
Anthropic just won it.
@robj3d3@levelsio@X Mine below 7% โฆ sometime when I realise my eye hurts itโs above 10% but also depends on the room brightness, i forget going back
@ann_nnng kinda similar, me everday until last last minute before going to work ... i give bunch or tasks and bugs to solve and keep my laptop open. i just need to test it when i'm back