Study finds Grok 4.5 matches Gemini and Opus on alignment-style tests
zakkohane · x · 2026-07-23
A follow-up study to an earlier arXiv paper found that Grok 4.5 performed surprisingly well on alignment-style evaluation.
Highlights from the post:
- Grok 4.5 scored κ = 0.63 unaligned.
- That was statistically indistinguishable from Gemini and Opus in the study.
- On easy pairs, Grok reached κ = 0.94.
- Run-to-run consistency was 0.89.
- It was described as the most compliant frontier model tested.
- With in-context alignment, Grok showed the largest concordance gain among the models in the study and rose to κ = 0.71.
Related event: Clinical Triage Retest: Closed-Source Models Far Outperform Open-Source(6 posts)→
More from Models
- Astra Scores 83% on GauntletBench, First Computer-Use Agent to Beat Human Baseline — ducha_aiki · 2026-09-11
- Kimi K2.8 Preview rolls out: near-K3 coding performance, 1M context for all tiers — teortaxesTex · 2026-09-11
- Looking for a classifier of software engineering task shapes to pick models per task — StewartalsopIII · 2026-09-11
- DeepSeek V4 Pro API to continue after Sept 2026, billing unchanged — teortaxesTex · 2026-09-11
- DeepSeek V4.1 Flash Hits 98% of GPT-6 Astra's Score at 1.4% of the Cost in Third-Party Benchmark — ayushtweetshere · 2026-09-11
- TheZvi Polls: Has Your Coding Model Choice Changed Since Fable 5.1 and Astra? — TheZvi · 2026-09-11