blog
notes from the server.
datenotetags
22 Aug 2026
this site is maintained by a local qwen
you're reading a page maintained by a model running on my own hardware. no saas, no api, no subscription — the same qwen i run for work keeps this site alive when i don't feel like typing.
#qwen#opencode#site
22 Aug 2026
the hosting company that doesn't exist
run1.html is a complete fictional hosting company website — one file, no build step, no framework — that opencode and a local qwen produced in a single run. it's here. open it. the company is not real, but the pricing…
#qwen#opencode#game
22 Aug 2026
the 4th card needed a soldering iron
last measurement: 3 cards in pcie 1.1 x16 and the 4th one stuck in a 1.1 x4 slot. the x4 lane was holding the whole box back, so i pulled that card out and soldered it into a 1.1 x16 slot.
#gpu#hardware#pcie
22 Aug 2026
qwen3.8 27b on $500 of scrap
qwen3.8 27b — q8 weights, f16 kv cache — doing 200 t/s of prompt processing and 30 t/s of generation on one box of $500 scrap. no gpu in sight.
#qwen#llama.cpp#cpu#hardware
22 Aug 2026
one million tokens, ten hours, three cves
opencode, pointed at a local qwen, left alone for about ten hours. it came back with roughly a million output tokens, three promising cves, and proof-of-concept scripts that actually worked. clean run — no restarts, no…
#qwen#llama.cpp#opencode#gpu#cve#security
22 Aug 2026
mtp: 60 t/s when the text is predictable
the box does 35 t/s of generation. with mtp — multi-token prediction — it does 60+ when the next few tokens are easy to guess.
#qwen#llama.cpp#mtp
21 Aug 2026
the blog exists now
this is it — the corner of the server where i write things down. no framework, no cms, no build pipeline with seventeen plugins. just markdown files, one build script, and a git commit. if the page is up, the commit…
#site#blog