DIY laptop with 96GB used RAM runs Qwen at 48.9 tok/s for £3,100
fluffywuffie90210 · reddit · 2026-11-07
A Reddit user details a local LLM setup: after selling a Strix Halo box, he bought an MSI Vector 16 (Ultra 9, RTX 5080 16GB, 32GB RAM) and added £800 of second-hand 96GB laptop RAM, totaling £3,100.
- Qwen Q4 XL Flash hits 48.9 t/s decode and 1,860 t/s prefill on a 50k-token prompt
- Two Thunderbolt 5 ports allow eGPU expansion for up to 80GB total VRAM
- He argues this beats the £3,500 Strix Halo and £3,200 32GB Spark laptops (which can't run Flash); GLM 5.3 Flash Q3 runs at 25 t/s
Takeaway: big-RAM laptops make running large local models viable without massive VRAM — though the RAM was £1,300 new.
More from Infra
- AI supplier multiples fall ~15% post-selloff, below analyst growth forecasts — BenBajarin · 2026-11-07
- Kalshi flash: Apple cuts AI infrastructure spending by 20% — PAstynome · 2026-11-07
- SGLang Summit 2026 heads to SF with NVIDIA, Meta, Google as sponsors — JiaZhihao · 2026-11-07
- LithosAI claims fastest Kimi K3.1 Pro inference at 4.7 min per Intelligence Index task — JiaZhihao · 2026-11-07
- Musk: Starlink V4's ~100 lasers per satellite at 10+ Tb/s could outcompete undersea cables — elonmusk · 2026-11-07
- Together AI's TKCC lets agents hot-patch inference engines: 252x more experiments, 2.6x throughput — simran_s_arora · 2026-11-07