Native Swift implementation of OpenJev, the Jev-compatible typed-decision server: yes/no, choice and score answers read from model probabilities, on Apple silicon Macs and in iPhone apps. Serves Verdict, Laya, DiffusionGemma and JevK5. Apache-2.0.

Axiocairn hosted MCP server: market briefs and stock model assessments with data dates and invalidation conditions.

按场景选模型的 8 维能力雷达图 · Scenario-based LLM capability radar with family-relative public benchmark scoring

王二讲Agent：AI Agent 面试题 / Agent八股，逐题图文详解+视频讲解，附 Remotion 视频工程

面向 AI 编程智能体的任务包工作流工具，用可验证的范围、证据和 Git 门禁管理长期改动。(Task-package workflow for AI coding agents with verifiable scope, evidence, and Git gates.)

Pulsar (formerly fastkernel): the fastest inference engine for Qwen3.8-27B and Qwen3.6-35B-A3B on Apple Silicon

Bilem Agent — программа для Windows: подключает 1С:Бухгалтерию для Казахстана к Claude, ChatGPT, Cursor и другим ИИ по MCP. Отвечает по данным 1С, печатает счета, доверенности и акты сверки; учёт не меняет. Open source (MIT). 🇰🇿

Agent serving trace from https://freeinference.org

DeepSeek-V4.1-Flash-1017 on one 256 GB M5 Ultra: 1M context, 70.7-104.9 tok/s decode with DSpark drafts, 1,739-1,867 tok/s cold prompts (8k-131k) with int8 x int8 prompt experts, a 1M prompt in 740 s, images, tools, JSON. TensorFold native Metal.

Run ChatGPT-style AI for free, offline and private on your own laptop. Beginner guide + a Windows tool that picks the right model for your hardware.

Catch LLM agents claiming actions they never took. Ledger + claim extraction + receipt for LangGraph.

Telegram bot that pulls a student's diary from nz.ua, finds the homework, locates the exercise in the textbook PDF (optionally a GDZ scan + OCR) and explains it via an LLM (Groq/Gemini). Ukrainian docs.

多模态工具调用 Agent 的后训练全链路：QLoRA SFT → RM/PPO → DPO → GRPO → AWQ 量化 → vLLM 部署 → 评测反哺课程，附 20+ 真实踩坑记录与原始日志

Language models running locally on hardware that has no business running them

Portable working context for AI coding agents: save your state once, and every agent (Claude Code, Codex, Cursor, Copilot, Gemini CLI) picks it up.

DAEDALUS: Bootstrapping Agent Memory from Self-Generated Tasks. Code, paper and blog post.

Claude Code in Neovim with Cursor-style hunk review: keep or revert each change Claude makes, review proposals before they are written, and run Claude full screen in zen mode. The up-to-date claudecode.nvim.

GPT-2 with zero floats: compress on a Mac, decompress on Windows/CUDA/Vulkan, byte-for-byte identical. Lossless LLM compression that doesn't break when the hardware changes. (alpha)

Open-source tool-calling platform for AI agents: 80+ integrations with OAuth, MCP server deployment, memory and agent flows.

AI toolkit for the Stellar/Soroban ecosystem: Python service + TypeScript SDK

Source-tracked catalog of abliterated, low-refusal, and community-reported open-weight and hosted language models

AIが書いた日本語を推敲するスキル yomiyasu を、スキルの大きさに上限がある環境向けに短くしたスキル（SKILL.md 18,000字以下）

Simulate communities of AI agents around a “what if…?” question. Native, local-first macOS app (SwiftUI) that runs on the Claude, ChatGPT/Codex or Ollama you already use. Free and open source.

Portable offline AI assistant for SAP/ABAP learning, code drafting, debugging, and development planning. Runs locally with lightweight models, streaming responses, personal note retrieval, and optional GPU acceleration.

DeepSeek Harness 官方通道破甲预设 · complete:true 整段替换 · 自带工具说明书 · 不改宿主 · 升级不丢

Real research papers, rewritten in plain words by small open models we trained. Live at sciencemadereadable.com

Claude Code plugin & skill that writes ad headlines by following an ad agency workflow (planner, CD, creative teams, client review) / 광고 대행사 작업 흐름을 따라 헤드라인을 쓰는 카피라이팅 플러그인

🎯 Free & open-source AI interview copilot. Real-time answers from voice + screenshots for coding and job interviews. Works with Groq, Claude, Gemini, GPT, DeepSeek, Ollama. A free alternative to Cluely & Interview Coder.

Self-hosted AI workbench that turns requirements into PRDs, technical specs, test cases, estimates, and interactive prototypes with Hermes Agent.

Native OpenCL inference engine for Intel N100-class SoCs: EmbeddingGemma-2, LiquidAI d1-omni and LFM2.5 chat in one C++ binary, on the iGPU. Measured ~2x llama.cpp's own Vulkan backend on the same N100 (2.1-2.3x); OpenAI-compatible server, CPU fallback, bit-exact batch invariance.