In-context alignment mainly helps frontier models on the hard pairs
zakkohane · x · 2026-07-23
Follow-up from the same study: the big movement is happening on the hard pairs.
- The chart shows that in-context alignment helps the frontier models much more on hard cases than on easy ones.
- Grok 4.5 rises from 0.28 to 0.46 on hard pairs.
- Llama 3.3 70B stays at -0.10 to 0.00, underscoring how weak the open-weight set was on the difficult part of the task.
Related event: Clinical Triage Retest: Closed-Source Models Far Outperform Open-Source(6 posts)→
More from Models
- BullshitBench update: GPT-6-Astra beats all prior OpenAI models but still trails Anthropic — scaling01 · 2026-09-11
- 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