Homelab on Rishi's Bloghttps://blog.bansalai.com/tags/homelab/Recent content in Homelab on Rishi's BlogHugoen-US&copy; 2026 Rishi BansalWed, 01 Jul 2026 00:00:00 +0000Running Local LLMs at Home: RTX 3060 vs 4090 vs 5090 on Qwen3 and Gemma 4https://blog.bansalai.com/posts/homelab-local-llm-gpu-benchmark/Wed, 01 Jul 2026 00:00:00 +0000https://blog.bansalai.com/posts/homelab-local-llm-gpu-benchmark/<p>A handful of consumer GPUs now sit at the center of serious homelab inference. The interesting question is no longer &ldquo;can I run a 30B-class model&rdquo; but &ldquo;which card should I buy for the workload I actually have.&rdquo; Below is a method-grounded look at three tiers — the RTX 3060 12GB, RTX 4090 24GB, and RTX 5090 32GB — running today&rsquo;s small-active-parameter mixture-of-experts models: <strong>Qwen3.5-35B-A3B</strong> and its newer sibling <strong>Qwen3.6-35B-A3B</strong>, <strong>Gemma 4 26B-A4B</strong>, and the dense <strong>Gemma 4 31B</strong>.</p>