having been a customer of Anthropic and Google at varying times, it's not surprising to me in the least.

As the companies sprint towards AGI as the goal the floor for acceptable customer service has never been lower. These two concepts are not unrelated.

Personally I use all of them all the time and chatgpt is still on top

Could you elaborate on your experience with the different ones? What you use them for and how they compare. Thanks

what do you use?

For conversational use, which is the main way these things are used, I personally found Claude to be the best.

Claude Sonnet is my favorite, despite occasionally going into absurd levels of enthusiasm.

Opus is... Very moody and ambiguous. Maybe that helps with complex or creative tasks. For conversational use I have found it to be a bit of a downer.

What are you smoking? Gemini and Claude beat chatgpt at every metric.

They don’t. GPT 5.2 and its variants are the best models right now.

Exactly. I don't understand how an article like this ignores the best models out there.

I doubt that gemini 3 cannot do it.

the thing is, I tried it and it took 10 seconds to import all settings from cursor. the moat for vscode clones is really small. i imagine people will jump a lot from clone to clone, like from model to model now.

sorry, but I don't understand you post. those links don't work.

You are the one who corrects america with usa?

it rivals a model that is obsolete? who is using openai deepresearch when there are so many better models out there?

almost no innovation? from transformers to alphago to the quantum stuff.

A bit like Xerox, Bell, Kodak or IBM back in the day then?

the reranker is a cross encoder that sees the docs and the query at the same time. What you normally do is you generating embeddings ahead of time, independent of the prompt used, calculate cosine similarity with the prompt, select the top-k best chunks that match the prompt and only then use a reranker to sort them.

embeddings are a lossy compression, so if you feed the chunks with the prompt at the same time, the results are better. But you can't do this for your whole db, that's why the filtering with cosine similarity at the beginning.