2026-11-09 · America/Los_Angeles · 社区动态 · #5

Stratoscope: Monitor local LLM token speed, cache hits & context usage live in your menu bar (Mac/Windows, MIT)

Stratoscope is a small tool for monitoring local LLM inference without having to open another window. It sits in your Mac menu bar or Windows taskbar and shows live prefill/decode tok/s, prompt cache hits, context fill, and RAM/GPU/CPU usage. Currently works with Strata, but each inference engine is a small adapter, so others are just one prompt away... Free and open source: https://github.com/yosephyne/stratoscope Short video here . Contributions and suggestions for other inference engines are welcome!

热度 46.0 / 100；排名与评分保留该期记录。