Opus 4.5 completes SWE tasks that would take a human ~4hr 49mins on average successfully 50% of the time.

However GPT-5.1-Codex-Max tops Opus 4.5 on the 80% success rate scoreboard at 32 mins vs. 27 mins

15h ago

Jobs

Take a 10-min AI interview to qualify for numerous real jobs auto-matched to your profile 🔑

+322 new users this month

Discover more

Curated from across

8,240

interviewed this week

Hire Top Engineers

Round1 finds you 10 great candidates in 48 hours!

roles listed this week

Take a 9-minute interview and auto-apply to top companies!

Ask a question on Grapevine.

Get the app on Android or iOS.