Qwen3.8

4 articles tagged with this topic

Qwen3.8 Crashes After 30 Hours—Small Model Done in 3: Architecture Beats Scale

Qwen3.8 took 30 hours and failed 16 cases; Muse Glimmer finished in 3 hours with higher implicit-knowledge scores. Architecture beats parameter count.

3d ago2 min read

AlibabaQwen3.8

Alibaba packs Opus-class AI into 24GB GPUs; open-source local LLMs truly work

Qwen3.8-27B hit Hugging Face trending #1 in 48 hours; Cline made it default in 4 days. Open-source LLMs hit a usable threshold for the first time.

Aug 222 min read

Qwen3.8Qwen3.6

Qwen3.8 27B Falters on 12GB VRAM: Older MoE Wins

Reddit test on a 12GB VRAM laptop: Qwen3.8 27B's 2-bit version misses classic logic puzzles, 3-bit stutters at 7.5 chars/sec. Older Qwen3.6 MoE wins.

Aug 162 min read

Qwen3.8Alibaba

Qwen3.8-27B Test: Top Reasoning Mode Is 7x Slower for Just 10% Better Output

Alibaba's Qwen adds three reasoning tiers. Test: top tier takes 6.4x longer but scores only 10% better. AI's diminishing returns are now visible.

Aug 162 min read