Grok 4.7 and Claude Opus 5.5 both score zero on CancerBench
iScienceLuvr · x · 2026-09-29
iScienceLuvr added the two newly released frontier models — Grok 4.7 and Claude Opus 5.5 — to his CancerBench benchmark, and they tied for first and last place with a score of zero. He stresses he isn't bearish on AI curing cancer; quite the opposite, he thinks far too little effort is going into applying frontier AI to cancer and hopes these posts draw attention so labs start saturating the benchmark soon.
More from Fun
- Why does Anthropic have so many interview rounds? A viral AI-circle joke — menhguin · 2026-09-29
- Conitzer's latest funny AI fail: 'the enemy of my enemy is my friend' — conitzer · 2026-09-29
- "How did he deploy Facebook without Vercel or GitHub?" — Paimaamu · 2026-09-29
- Browser-to-Browser Multiplayer Works, Even Unreal Tournament '99 Exists — banteg · 2026-09-29
- Dev jokes that adding yourself to the docker group is basically installing a rootkit — haydendevs · 2026-09-29
- Every academic has two citation counts: the one they don't care about publicly and the one they know exactly — prof_kamilov · 2026-09-29