Head-to-head on one 5090: Qwen 3.8 27B beats Flash Next, 86.15% vs 79.22% on LiveCodeBench
Fragrant_Scale6456 · reddit · 2026-10-31
A Reddit user tested Qwen 3.8 27B against Flash Next (QFN) in each model's best configuration on a 5090 32GB + 64GB DDR rig: QFN on basalt nvfp4 with 256K int8 KV (one slot), 27B on nvfp4 quasar QAT with a 570K KV pool (four concurrent slots).
Results were decisive for 27B:
- BFCL item-weighted: 81.51% vs 78.19%; LiveCodeBench pass@1: 86.15% vs 79.22% (and QFN logged 27 empty answers counted as fails)
- 27B dominates long context and hard problems: multiturnbase 64% vs 56%, multiturnlongcontext 60.5% vs 54.5%, LCB hard 75% vs 59%
- QFN only edges short single-turn categories (simplepython 92.5% vs 89.75%)
Author's verdict: matching his vibes-based impression — four concurrent agents with a huge KV pool is too good to give up for QFN.
More from Models
- Dev tasks Claude Fable 5.1, GPT Astra and Grok 4 with overnight long-running job comparison — therealdanvega · 2026-10-31
- DeepSeek V4.1 Flash non-reasoning hits ~1,486 tokens/s on Artificial Analysis at Intelligence Index 18.5 — teortaxesTex · 2026-10-31
- Qwen4 with HC architecture could widen the Q4 vs Q6 quantization gap significantly — Ok-Shower7286 · 2026-10-31
- Apple Intelligence confirmed for China: powered by Alibaba's Qwen, arriving with iOS 27.3 in December — op7418 · 2026-10-31
- Developer pushes back: 'speech-to-speech' should mean native audio turn-based LLMs only — pbbakkum · 2026-10-31
- Ex-OpenAI researcher: biological-compute hype isn't a real ML breakthrough — willdepue · 2026-10-31