Gamer Builds Local LLM Home Lab with Four 16GB GPUs
HippEMechE · reddit · 2026-07-07
A Reddit user shared a local LLM home lab built on four 16GB GPUs. Using main-slot bifurcation and PCIe 3.0 x1 risers, it runs two llama.cpp instances and deploys Qwen 3.6 with speculative decoding (q40, 150k context per instance), achieving around 1000 tok/s prompt processing and 45-60 tok/s generation.
They built a backend using opencode to manage llama.cpp and token counting, estimating they have already saved about $60 in API fees.
More from Infra
- Meta approves nonvoting Class C shares as 2026 capex guidance hits $135B–$145B — thione · 2026-10-26
- Unsloth Ships MiniMax-M3.1-Flash Quants: 2-bit Retains 91.5% Accuracy, Runs in 150GB RAM — danielhanchen · 2026-10-26
- Engineer with thousands of fabricated devices pours cold water on photonic crystal modulator yields — jwt0625 · 2026-10-26
- Anthropic S-1 reveals 48 cents of compute cost per dollar of Q2 revenue — nmasc_ · 2026-10-26
- Persistent AI joins UR network to route private AI inference with encrypted, IP-hidden traffic — 0xSammy · 2026-10-26
- Anthropic's disclosed compute commitments: $114.5B Google, $161.2B Broadcom top the list — BenBajarin · 2026-10-26