Qwen 3.8's structurally correct but nonsensical replies baffle Reddit users
TastesLikeOwlbear · reddit · 2026-09-17
A Reddit user complains that Qwen 3.8 frequently emits sentences that are structurally correct but make no actual sense — reminiscent of a Mitch Herdberg bit about tricky survey questions. Examples: "Are you trying to start a new project or create something from scratch?" and an RPG attempt: "I'll trade you a piece of information I haven't spent yet."
The author says he understands optimizing for coding and agentic use, but these models are getting very difficult to talk to.
More from Models
- Claude upgrade-page traffic rebounded to 17.6M in September, near May's 18.6M peak — firstadopter · 2026-11-02
- Open-source GLiNER-PII scrubs 13 identifiers from a discharge note in 550ms on GPU — MaziyarPanahi · 2026-11-02
- OpenAI Dots user says bloat and broken infra drove him to try newcomer Muse — Koala_Confused · 2026-11-02
- Fable 5.5 reportedly one-shots animation in pure code as Anthropic's coding bet pays off — Hesamation · 2026-11-02
- One agent goal has been running for 4 days, burning 30K tokens — CtrlAltDwayne · 2026-11-02
- Daily AI Brief: Gemini 4 Carbon spotted on Arena, DeepSeek open-sources coding agent — testingcatalog · 2026-11-02