Qwen 3.8 27B Quantization Tests: 200K Context on 16GB VRAM

Reddit users report running the quantized Qwen 3.8 27B model on consumer GPUs, achieving up to 200K context on 16GB VRAM and 50 tok/s with 100K context, though some note excessive thinking can bloat context.

2026-08-28 ~ 2026-08-29 · 4 related posts

• Qwen 3.8 27B quantization achieves 200k context on 16GB VRAM — abskvrm · 2026-08-28
• Running 27B model on 12GB VRAM: Qwen 3.8 quantization benchmark — Square_Light1441 · 2026-08-28
• Is Qwen 3.8 27B at Q2 quantization still usable? A 16GB owner asks — Effective_Head_5020 · 2026-08-29
• Qwen3.8 27B runs at 50 tok/s with 100k context on 16GB GPU — qaf23 · 2026-08-29