Hacker News new | past | comments | ask | show | jobs | submit
Many people rightfully compare these models on code generation and other long term tasks. Very few people are using these for researching things and analyzing what they have found so far from the sources, per my observation (I might be wrong, though).

On that latter category, I find Gemini superior to Claude for many tasks. Of course Google has an advantage when it comes to reaching places in the internet via its index and things like YouTube, but Gemini is really more useful than people give credit for the tasks I use it for.

Agreed. I enjoyed Argon and Carbon for general reasoning especially when it's used for context outside the US e.g. in Southeast Asia.

Claude's response is often very textbook and ignores the reality on the ground ( corruption, non enforcement of laws and regulations, shakedown from local magnates / police, etc).

Are you talking about AI Mode, or Gemini app with search enabled? I'm using AI Mode 95% of the time, and few months ago had much worse experience with Gemini. Working in AI deep research space, also agree that Google is SOTA right now.
loading story #50571435