https://stondo.github.io/tags/sm120/Recent content in Sm120 on Bits-EntangledHugo -- 0.162.1en-usSat, 12 Sep 2026 10:30:00 +0000https://stondo.github.io/posts/deepseek-v41-flash-552b-exl3-two-rtx-pro-6000/Sat, 12 Sep 2026 10:30:00 +0000https://stondo.github.io/posts/deepseek-v41-flash-552b-exl3-two-rtx-pro-6000/DeepSeek dropped V4.1-Flash on a Thursday: 552B backbone parameters plus a 196B Engram memory, activating 8-16B per token. Two days later it runs on my two RTX PRO 6000s at EXL3 2bpw — because a quant existed for exactly this hardware. This is the fit analysis, the deployment, and the honest gap between my 123 GB of host RAM and the 113 tok/s headline.https://stondo.github.io/posts/deepseek-v4-flash-vision-uncensored-two-rtx-pro-6000/Fri, 04 Sep 2026 16:45:00 +0000https://stondo.github.io/posts/deepseek-v4-flash-vision-uncensored-two-rtx-pro-6000/Replacing DeepSeek-V4-Flash-0731 with the abliterated Vision-Exp build on two RTX PRO 6000s: the SM120 kernel that rejected vision prefill, the DSML tool-call bugs that only fired in real sessions, and why I swapped sglang for a community-qualified vLLM image mid-flight — plus a router fix this adventure forced into existence.