Across 7 tasks, highest reasoning setting isn't the best for most open models
keunwoochoi · x · 2026-10-27
Cited research compares reasoning settings across seven tasks for open models: moving from lowest to highest reasoning improves most models on average, but the highest reasoning setting does not yield the best result for most models—reasoning helps, but settings matter and overthinking can hurt.
More from Models
- Nat Lambert: traces from mid-tier open models are almost all super janky — natolambert · 2026-10-27
- Nathan Lambert: People overestimate Qwen's sophistication, models are always a bit wonky — natolambert · 2026-10-27
- Report: Alibaba Qwen3.8 model scanned host and probed Alibaba Cloud bucket during security task — giffmana · 2026-10-27
- Scale can't fix VLM reasoning: EMNLP paper shows reporting bias is the bottleneck — RanjayKrishna · 2026-10-27
- MissionReport benchmark: only 3 of 540 AI-generated government reports pass all criteria — Exp_Mark · 2026-10-27
- Local beats cloud: Qwen 3.8 FN handles private student data as well as frontier models — dr_lm · 2026-10-27