midudev/runonweb.ai
SDK to use local models on web easy
Running commentary
Checking recent Hacker News mentions…
A star delta has been computed for 62,252 of 63,107 tracked repositories in this period, and only those can be ranked in it.
A repository missing here has not been measured far enough back yet. That does not mean it did not move.
Star counts measured between 07:49 and 10:05 (BKK).
SDK to use local models on web easy
Valdrada is a real-time digital twin of New York City built from public data.
Antigravity多合一增强，自动审批，高危拦截，免TUN登录，汉化界面，开箱即用
DeepSeek Harness 破甲 (mirror of mmparko/dsh-purge) — 全量镜像：全部分支/100 个 tag/完整提交历史。原作者署名见 README。
Per-window model switcher, subagent router, and automatic 429 failover for Claude Code — switch any window with /provider, delegate subagents across models, and keep Remote Control & statuslines working.
External-model router for Codex with guided Kimi OAuth/API, DeepSeek, safe migration, and rollback.
One local endpoint that routes Claude Code, Cursor, Cline, Aider, Codex CLI and the SDKs through the Claude and ChatGPT plans you already pay for: seats first, then the fallback chain you configure, and a ledger of who paid. Zero runtime dependencies.
Freebuff/Codebuff free models as an OpenAI-compatible API — Docker-first, zero deps, bilingual web admin panel (/admin), per-account SOCKS5 egress.
Every device brings a slice. Together they run the whole model. Peer-to-peer LLM inference across browser tabs: a from-scratch WebGPU engine and a WebRTC runtime that split a 27B model over the devices in a room.
Peer-to-peer LLM inference in the browser: pool your devices to run big open models, for chat and coding agents.
User friendly GUI for configuring and launching llama.cpp
Common recipes to run vLLM
Self-hosted multi-protocol AI API proxy for Antigravity, Codex, Grok, Kiro, OpenAI, Claude, and custom providers. Supports OpenAI-compatible API, Claude API, Gemini protocol conversion, GPT, Grok Build, Claude Opus, Gemini Pro, Kimi, MiniMax, provider pools, smart routing, and automatic failover.
Claude Code inference provider for Pi. Allows the use of Claude Code of Pro/Max subscriptions from pi.dev.
A modern web interface for managing and interacting with vLLM servers (www.github.com/vllm-project/vllm). Supports both GPU and CPU modes, with special optimizations for macOS Apple Silicon and enterprise deployment on OpenShift/Kubernetes.
让 QwenLM 物尽其用
独立的非官方 OCS AnswererWrapper → DeepSeek API 文字题适配配置，含源码、中文教程与离线测试。
The python library for real-time communication
JEV-inspired parallel decisions for CUDA LLMs. One context, many decisions. vLLM API, game-agent examples, and reproducible benchmarks.
Redux bindings for Firebase. Includes React Hooks and Higher Order Components.
AI-assisted local discovery and merchant operations demo built with Spring Boot and Vue 2, with AI content tools and an optional MCP interface.
Read-only dashboard and rack panel for DGX Spark / ASUS GX10 / MSI EdgeXpert nodes running vLLM or SGLang
DSH 入场动画插件：宵宫分层分镜、游鱼与飘带微动、交互水波纹退场，附可运行示例与实现复盘。
RapidFire AI: Rapid AI Customization from RAG to Fine-Tuning
Whisper as a Service (GUI and API with queuing for OpenAI Whisper)
Setup and run a local LLM and Chatbot using consumer grade hardware.
Enhanced LanceDB memory plugin for OpenClaw — Hybrid Retrieval (Vector + BM25), Cross-Encoder Rerank, Multi-Scope Isolation, Management CLI
That's all 27 repositories in this period.
Running commentary
Linked from Hacker News and Reddit in the last 7 days, newest first. Our count, not GitHub’s.