85.6 tok/s on a single RTX 5090: local Qwen3 27B with MTP speedup
comperr · reddit · 2026-09-26
A Reddit user reports running Qwen3.8:27B locally on a single RTX 5090, achieving 85.6 tok/s using MTP (multi-token prediction). A useful data point for anyone considering running 27B-class models on consumer hardware.
More from Infra
- Running Qwen2.5-Coder-3B on Android: Jupyter + Pydantic Evals on-device eval pipeline — alex_b3b · 2026-11-11
- 22% of Nvidia revenue now books to Singapore, but only ~1% is local buyers — Gregory_C_Allen · 2026-11-11
- Huawei's Kirin 9050 NPU claims offline 30B MoE on smartphones, +140% perf — AIFlow_ML · 2026-11-11
- Stratechery: Intel vs AMD's diverging earnings and Intel's need for an old-new path — Stratechery · 2026-11-11
- a16z Podcast: Mike Volpi on What the 1990s Internet Boom Teaches Us About the AI Buildout — a16z Podcast · 2026-11-11
- Alibaba files to list chip-design unit T-Head in Hong Kong to fund AI push — pstAsiatech · 2026-11-11