Observed arrival · 2026-10-18
Luna — a local AI assistant
A personal, locally run Qwen3 14B assistant, documented through GPU benchmarks, tool-routing notes, and reliability tests.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- People testing local Qwen inference on AMD GPUs
- Worth noticing
- The benchmark uses seven-run medians with a fixed seed; output quality was not compared.
Field notes
The page lays out a local llama.cpp pipeline from assistant logic to a Qwen3 14B Q4_K_M model on a 16 GB GPU, and links commands plus bench.py for the Vulkan/ROCm runs. Its reliability report counts 109 passes across 110 evaluated attempts (22 scenarios, five runs each); 15 attempts were skipped when preconditions failed, and the author says a cleaned-up run script was not rerun.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
○OpenPublic substance visible
✦PrettyNotable craft visible
●ProPolished or operationally mature
◎NicheUnusually specific use
◉HumanPersonal, local, civic, or handmade
One card from the complete issue