Qwen 3.5 Setup on Your Local Computer (Step-by-Step Guide)

Qwen 3.5 runs locally on LM Studio, but it flunks the car-wash test

The creator walks through installing Qwen 3.5 via LM Studio (cross-platform Windows/Mac), choosing the 9B parameter / 6 GB model, full GPU offload, and pulling it into a chat on a PC with an Nvidia 3060. Download speed and installation are literally a next, next, next click flow.

Hardware floor and ceiling

The host's rig: older PC, RTX 3060, and he still got 37 tokens/second on the 9B Qwen 3.5 build. Larger Qwen 3.5 checkpoints go up to 22 GB with partial GPU offload; the more RAM you have, the higher quality you can load up. Apple's own Mac Mini page was shown in Chinese first, then switched to English — the point being that paid Apple silicon = more headroom for bigger Qwen variants.

The sanity check that broke it

The recurring test: I have a car wash 50 meters away from me. Should I walk or drive? Qwen 3.5 got stuck reasoning for ~2 minutes, eventually produced a gaslight-y response about walking is usually the more logical choice, then generation failed. For contrast, the host notes Grok 3 said drive on the same prompt and he flags the failure as concerning.

Speed is real, intelligence is not

Throughput on the 3060 is genuinely good: the model fought for 3.5 seconds on a name question and answered a keto meal plan request very fast. But the creator's bottom line is I'm still gonna stick with Opus and labels local Qwen play at your own risk — fine for learning apps or I'm on keto what make make me a meal plan-style basic tasks, not for decisions you actually care about.

Disable thinking mode

One concrete fix surfaced during the test: toggling disable think made Qwen 3.5 finally answer the car-wash prompt without looping. The host's takeaway — just disable thing and I think you're in good shape — is the only actionable workaround he offers before signing off with play at your own risk, enjoy.