Unsloth: Local UI for Running and Training LLMs
A local interface to run and train LLMs and diffusion models. It supports GGUF, MLX, Qwen3.8, DeepSeek-V4, Gemma 4 and FLUX.
Image: GitHub
The weekly · Week 33 · 10 Aug – 16 Aug 2026
This week's items cover running and training models locally, routing LLM traffic across providers, and tools for coding agents such as plugins and diagrams.
8 picks
A local interface to run and train LLMs and diffusion models. It supports GGUF, MLX, Qwen3.8, DeepSeek-V4, Gemma 4 and FLUX.
Image: GitHub
Editorial diagram design for Claude Code, Codex, GitHub Copilot, Factory Droid and Pi. It offers 42 diagram types as self-contained HTML and SVG, without Mermaid.
Image: GitHub
Switchyard lets LLM applications route traffic across models and providers while keeping native OpenAI and Anthropic API compatibility, for model selection, benchmarking and cost tuning.
Image: GitHub
A GitHub repository named awesome-deepseek-agent from deepseek-ai. Its description is not yet available.
Image: GitHub
Needle2 is an 8 to 29 MB model for phones, wearables, robots and smart home devices. The project says it beats models 10x its size on mobile tool calls.
The Cursor plugin specification together with the official plugins, which developers can use as a reference when building their own.
Image: GitHub
A step-by-step guide to running managed AI inference on Google Cloud, with architecture, code, deployment and security explained.
Nightcrawler is a local AI pentesting agent that runs on a smartphone, shared as a GitHub project.
Picked from public sources (GitHub, Hacker News, DEV, Lobsters, Reddit, Product Hunt) by our daily scout. Each summary uses only what the source itself says; GitHub stars are as checked on the date shown.