Discussion on Qwen3.8-27B Reasoning Budget Configuration and llama.cpp Implementation
Thin_Pollution8843 · reddit · 2026-08-24
A user discusses issues with the reasoning budget feature in Qwen3.8-27B, noting that the model sometimes thinks too deeply and hits the 128k output token limit. The post compares the official Qwen thinkingbudget configuration with the llama.cpp --reasoning-budget parameter implementation (which includes a stop message) versus a pi plugin implementation, seeking a more generic solution.
More from coding & agent
- raylib 6.2 ships with 300+ improvements and 100 new contributors, 20 new API functions — jsuarez · 2026-10-16
- "Vibecoding" is as misleading a label as "AI" and "SI," dev complains — eschadiol · 2026-10-16
- Doc2Harness: New benchmark finds best coding agent passes only 53% across 329 cases — _akhaliq · 2026-10-16
- Perplexity open-sources SPACE, the Firecracker microVM sandbox behind Perplexity Computer — perplexity_ai · 2026-10-16
- Blackstone goes all-in on Factory for autonomous software development — matanSF · 2026-10-16
- Longtime Windsurf user cancels $10/month legacy plan, citing AI subscription fatigue — tristanbob · 2026-10-16