Thriving

TS

CS

↑ 51 ?

CookLLM Crvesoft/CookLLM

A lightweight launcher for llama.cpp / Ninfer / KVMem: model hub, multi-engine management, parameter presets, one-click start/stop, performance monitoring, community model exploration and downloads, and a built-in chat. A modern GUI launcher for llama.cpp / Ninfer / KVMem to manage models—multi-engine hub, presets, one-click server, live metrics, downloader & chat UI.

HEALTH ? 89 / Healthy

// solo builder// hidden gem// release machine ?

// readme

🦙 CookLLM

Lightweight manager for llama.cpp / Ninfer / KVMem: a Windows desktop manager for GGUF / Ninfer models. It includes a built-in model library, runtime parameter templates, one-click start/stop, GPU status monitoring, live run logs, and a built-in chat web page, so you can manage local models in one place.

✅ Changelog

v0.4.2

🚀 New charts and card improvements

New charts on the analysis page: added live Decode / Prefill charts, plus TTFT (time to first token) and MTP acceptance rate cards (automatically hidden when MTP is not enabled).
Card animation improvements: model cards use a unified amber breathing state while starting; fixed the start button not responding promptly, and readiness is now determined by an HTTP probe.

🛠️ Analysis sampling fixes and tuning

Sampling fix: captures tail bursts that end within the sampling interval; the Prefill readout is limited to a 60-second window.
Prefill counting: the llama.cpp Prefill count now uses n_prompt_tokens_processed.

v0.4.1

✨ Polishing UI details

Dropdown lists: optimized the top-bar dropdowns, aligned to equal-width pills with single-line…