Linus runs Qwen3 27B in LM Studio, gets roasted for skipping vLLM's 5x speed
victormustar · x · 2026-11-03
In the latest Linus Tech Tips video, Linus runs Qwen3 27B GGUF (Q4KM) with pipeline parallelism in LM Studio and complains it's slow. @Youssofal argues he could get 5x better speed and intelligence with vLLM-based tooling — a one-time setup. victormustar shares it with the classic "miner walking away from diamonds" meme.
More from Fun
- Flower Labs' YC-backed federated learning site mocked as AI-generated slop — teortaxesTex · 2026-11-03
- Claude Code replicates Doppler-based location leak: pinpointing a webcam from audio alone — flngr · 2026-11-03
- Hugging Face dev open-sources the model he uses every day — lhoestq · 2026-11-03
- Google AI Mode keeps leaking image carousel metadata as plain text — One-Nobody639 · 2026-11-03
- Training neural nets inside LLMs: XOR MLP goes 53% to 98.8% in 40 steps on simulated QwenGPUs — zeeshanp_ · 2026-11-03
- Tiny latent diffusion model trained inside LLMs generates MNIST digits, 76% recognized — zeeshanp_ · 2026-11-03