Qwen 3.8 vs 3.6: Low reasoning mode loops less
Lair98 · reddit · 2026-08-22
User reports that Qwen 3.8 loops significantly less than 3.6 when reasoning is set to 'low', even on 3-bit quantization.
Key findings:
- The model defaults to xhigh; users must specify 'low' to reduce looping.
- A new parameter preservethinking allows keeping/discarding the reasoning process after each turn. Activating this prevents the model from re-reasoning the same content.
- Real-world testing shows 3.8 'low' mode is an improvement over 3.6, though not perfect.
More from Models
- Claude Opus 5.5 effort levels tested: low reasoning iterates 9x faster than max — tom_doerr · 2026-10-16
- Muse Spark Open runs locally on a Mac M3 Ultra via llama.cpp: frontier model on desktop — pcuenq · 2026-10-16
- OpenAI ships GPT-Live-1.1: new voices in six languages, fewer confirmations — OpenAI · 2026-10-16
- Muse Spark model appears on LMArena, testable in Battle and Agent modes — arena · 2026-10-16
- Running a multi-model UI design arena: Paseo orchestrates four design drafts per model — vista8 · 2026-10-16
- Ilya Sutskever questions Meta license clause taxing MaaS providers above $20M revenue — ilyasu · 2026-10-16