Quesma Blog company blog
Benchmarking Qwen3.8 27B quantizations: 4-bit holds up, 1-bit collapses
I test Unsloth GGUFs of Qwen3.8 27B (Q4_K_M, UD-Q2_K_XL, UD-IQ1_S) with llama.cpp on GPQA Diamond, IFBench, Terminal-Bench 2.1. Q4_K_M matches BF16 and fits an RTX 4090.
- Points
- 286
- HN comments
- 139
- Submissions
- 1
- Cited by
- 0 commenters
AI & Machine LearningTools & Productivity2026
On Hacker News
| Submitted | Title on HN | Points | Comments |
|---|---|---|---|
| Sep 8, 2026 | Benchmarking Qwen3.8 27B quantizations: 4-bit holds up, 1-bit collapses by stared | 286 | 139 |
