麦门封神榜|你在麦门算老几?领取你的麦门专属身份
GLM-OCR: Accurate × Fast × Comprehensive
GLM-4.6V/4.5V/4.1V-Thinking: Towards Versatile Multimodal Reasoning with Scalable Reinforcement Learning
An open-sourced end-to-end VLM-based GUI Agent
支持OneBot(QQ)、OneBotV12、QQ频道、微信、KOOK(开黑啦)、Telegram(电报)、FeiShu(飞书)、DoDo、Villa(米游社大别野)、Discord的全功能HoshinoBot/NoneBot2/Koishi/yunzai/ZeroBot/AstrBot的原神机器人插件
An all-in-one, 100% local AI video, image, and music studio. Director mode plans full music videos and short films from a single prompt. Built on the WanGP pipeline. Install via Pinokio.
🪐 HelixWorld: real-time interactive audio-visual world model.
Allow your 🦞 bot to Shout, Speak, with "human" vibe
Real-time, interactive spatial audio generation for videos and world models.
ZCode Plugins Marketplace - official built-in and community plugins for ZCode
Fast, concurrent inference for YuE2. Same model and recipe: 1.68× faster per song, 3.31× more songs per GPU.
Automate the ZCode free Start-Plan claim (z.ai token/cookies -> ZCode JWT -> Start Plan) and wire GLM into 9Router as a glm provider pool. Web UI console + CLI. Reverse-engineered from ZCode 3.14.4.
Code for Large-Discovery-Model (LDM)
Turn complex sources into evidence-traceable, layout-safe, natively editable McKinsey-style PowerPoint — 45 semantic page contracts and verified chart data.
Independent, evidence-based map of when TypeSafe's Jev actually holds up vs. breaks down — real API-call receipts, not a leaderboard. 中文為主的雙語 repo。