Beautiful dashboard for running local LLMs 😍
🦙llamastash — A TUI dashboard for managing local AI models
💯 Discover, download, launch & monitor models with llama.cpp, vLLM, SGLang & more
🦀 Written in Rust & built with @ratatui.rs
⭐ GitHub: github.com/llamastash/l...
#rustlang #ratatui #tui
Beautiful dashboard for running local LLMs 😍 🦙llamastash — A TUI dashboard for managing local AI models 💯 Discover, download, launch & monitor models with llama.cpp, vLLM, SGLang & more 🦀 Written in Rust & built with @ratatui.rs ⭐ GitHub: github.com/llamastash/l... #rustlang #ratatui #tui
0 vue
Publié le 27/10/2026 11:41
0:38
Beautiful dashboard for running local LLMs 😍
🦙llamastash — A TUI dashboard for managing local AI models
💯 Discover, download, launch & monitor models with llama.cpp, vLLM, SGLang & more
🦀 Written in Rust & built with @ratatui.rs
⭐ GitHub: github.com/llamastash/l...
#rustlang #ratatui #tui
2 commentaires
?
A per-model VRAM readout is the useful part. Worth showing weights vs KV cache separately: a 32B at Q4_K_M is ~19 GB of weights, but 32K context at fp16 KV adds several GB more, and that is what silently pushes layers to CPU and drops decode from ~30 tok/s to single digits.
A TUI that shows VRAM per model at a glance would save me so many nvidia-smi tabs. Does it show the CPU/GPU split when a model spills over?