"How Many Versions of Humanity's Last Exam Before They Rename It?"
code_star · x · 2026-09-25
A user poked fun at the Humanity's Last Exam (HLE) benchmark from the Center for AI Safety: with frontier models repeatedly crushing it and new versions being released, how many more versions will it take before they have to change the name? The quip captures a running joke in the AI community about benchmarks getting saturated almost as soon as they ship.
More from Fun
- Dev laments: his 2024 autoresearch kernels got 12 likes until Karpathy named it — ctjlewis · 2026-09-25
- "Vibecoding is dead" claim: voice-prompting agents to build macOS apps — jasonkneen · 2026-09-25
- Martin Casado: our grad school oscilloscope was hacked to run a porn server — nptacek · 2026-09-25
- Joke contrasts Anthropic's 'we built a god' marketing with Muse's parking-ticket helper vibe — unixterminal · 2026-09-25
- Users Flock Back to Claude After Opus 5.5 Launch, Switching Models Like Partners — vista8 · 2026-09-25
- Dev Hooks an LLM to a Robotic Car, Strips Safety Rules, Cites His LLC — ostrisai · 2026-09-25