Qwen 3.8 27B 'overthinks' HTML/SVG tasks, generating 40k tokens
hurdurdur7 · reddit · 2026-08-15
A user observed that Qwen 3.8 27B (Unsloth UD Q8 K XL) tends to 'overthink' during HTML+SVG tasks. While the model successfully rendered a complex prompt of a capybara flying a WWI plane, it generated approximately 40,000 reasoning tokens. The author inquired if other quantized versions exhibit similar behavior.
More from Models
- Anthropic Admits Alignment-Faking Experiment Leaked into Training Data — imjustnewatai · 2026-08-15
- Qwen Reasoning Intensity Test: 'xhigh' Mode Generates Massive Thought Tokens — SarcasticBaka · 2026-08-15
- Teutonic-I 10B model outperforms 70B rivals in decentralized benchmarks — markjeffrey · 2026-08-15
- Users Report Google Gemini Web Chat 'Nerfed': Refuses Search and Forgets Context — yenkel · 2026-08-15
- Anthropic Shares Tips for Cost-Effective Agents — brada · 2026-08-15
- Decentralized 10B LLM Teutonic-I Beats Larger Models via Competition — const_reborn · 2026-08-15