π¬Community
Saturday, January 17, 2026
Claude Code Performance
Users are comparing Claude Code's performance on benchmarks and noting its occasional refusals on certain topics and its superiority to Codex and GPT.
2 tweetsβ’2 engagements
@oznova_Oz
I've been running Claude Code against some benchmarks including LAB-Bench, BixBench and HLE. It'll frequently refuse on bio questions, never on anything else
β₯ 2
@Ashrya3Ashrya Agrawal
After ~7 months, gave codex(5.2-xhigh) and GPT a try due to the recent hype. Iβm surprised itβs still not even close to the Claude code v1 we had 6 months back.
Part of newsletter