Benchmark: Qwen 27B 3.8 outperforms 3.6 in web design tasks
ShadyShroomz · reddit · 2026-08-15
A user ran a personal benchmark comparing Qwen 3.8-27B against version 3.6 on web design tasks. The models were run at full bf16 precision. Results indicated that the new 3.8 version produced superior design aesthetics. Comparison images and detailed benchmark scores are available via provided links.
Related event: Qwen 3.8 27B Released as Community Tests Performance and Speed(4 posts)→
More from Models
- willcb: Opus 5.5 doesn't smell bigger, it smells like they fixed the data — willcb · 2026-10-16
- Unverified claim: Apple and Meta's 816B-parameter Muse Spark Open runs on two Mac Studios — TansuYegen · 2026-10-16
- Gemini 4 model card: model eliminated months of trial-and-error on bioweapons research — Puzzleheaded-King584 · 2026-10-16
- Every AI picks 7: Anthropic engineer's rule for escaping model taste convergence — rubenhassid · 2026-10-16
- Brain teaser: strip all attention, MLP and embedding matrices from a Llama 3 — how many trainable params remain? — Sauers_ · 2026-10-16
- Lean v4.35.0 ships verified kernels — one written and proven consistent by Claude — geoffreyirving · 2026-10-16