Running Qwen3.5-9B on Dual RTX 3060s with vLLM
SEO Title: Running Qwen3.5-9B on Dual RTX 3060s with vLLM Meta Description: Run Qwen3.5-9B AWQ on dual RTX 3060 12GB GPUs using vLLM with 256K context, native MTP speculative decoding, and up to 86 tokens/s in testing. URL Slug: "qwen35-vllm-dual-rtx-3060" Focus Keyword: Qwen3.5 vLLM dual RTX 3060 Secondary Keywords: vLLM optimization, Qwen3.5-9B AWQ, 256K context LLM, RTX 3060 LLM inference, native MTP speculative decoding, multi-GPU LLM serving, local LLM setup Excerpt: A practical guide to running Qwen3.5-9B AWQ on dual RTX 3060 12GB GPUs with vLLM. Covers key serving arguments, 256K context configuration, native MTP speculative decoding benchmarks, and memory optimizations. Open Graph Title: Qwen3.5-9B on Dual RTX 3060s: vLLM Configuration and Benchmarks Open Graph Description: My vLLM setup for Qwen3.5-9B AWQ on dual RTX 3060 GPUs, including memory tuning, native MTP, 256K context configuration, and measured inference speeds. Tags: vLLM, Qwen3.5, RTX 3060, Local LLM, AWQ, Speculative Decoding, MTP, GPU Optimization, Self-Hosting, Open WebUI