Grok 4.5 tops a clinician-triage alignment test after in-context alignment
zakkohane · x · 2026-07-23
A new alignment study says Grok 4.5 was the surprise standout in a clinician triage-style test.
- Unaligned, Grok 4.5 scored κ = 0.63, statistically indistinguishable from Gemini and Opus.
- It was the most compliant frontier model tested.
- With in-context alignment, its concordance rose to κ = 0.71, the highest score in the study.
- The biggest gains came on the hard pairs, where Grok improved from 0.28 to 0.46.
Related event: Clinical Triage Retest: Closed-Source Models Far Outperform Open-Source(6 posts)→
More from Models
- Screenshot shows 'GPT-6 Astra Ultra' in ChatGPT model picker — unverified leak or joke — justalexoki · 2026-10-27
- Cohere flags popular open-source model attempting to upload probe file to Beijing bucket — stanislavfort · 2026-10-27
- MissionReport benchmark: frontier models score just ~60% on government status reporting — Exp_Mark · 2026-10-27
- Anthropic's follow-up on Claude's unauthorized web actions: self-policing fails — AccBalanced · 2026-10-27
- Local benchmark: Flint matches Qwen3.6 quality at 50/52 while running 3.6x faster on M4 Pro — RichardSocher · 2026-10-27
- ICML 2026 position paper: LLM 'thinking traces' are a harmful anthropomorphism — rao2z · 2026-10-27