https://blog.lightsoutlabs.ai/ Engineering notes from Lights Out Labs. Fri, 21 Aug 2026 00:00:00 +0000 https://blog.lightsoutlabs.ai/posts/running-qwen38-27b-on-one-rtx-3090/ https://blog.lightsoutlabs.ai/posts/running-qwen38-27b-on-one-rtx-3090/ Local Qwen3.8-27B peaks at 81.4 tokens/s with a few perf adjustments. Fri, 21 Aug 2026 00:00:00 +0000 https://blog.lightsoutlabs.ai/posts/muse-glimmer-first-impressions/ https://blog.lightsoutlabs.ai/posts/muse-glimmer-first-impressions/ Muse Glimmer is a flop: chatty, prone to hallucinations, bad at tool calling, slower than Qwen and only half of Qwen 3.6's context. Tue, 11 Aug 2026 00:00:00 +0000