Open Source LLM Projects
Discover 263 open source LLM repositories from GitHub, each with an analysis of what it does, key features, use cases and alternatives. LLM projects here are most often combined with Python, AI and AI Agents. Last updated October 3, 2026.
263 repositories · updated October 3, 2026
awesome-ai-agents: Discover AI Agent Tools and Research
A community-curated directory of AI agent frameworks, platforms, infrastructure, evaluation tools, safety resources, papers, and learning materials. Use it to survey the ecosystem and find options for a particular agent-building need.
A-MEM: Self-Evolving Memory for Coding Agents
A-MEM is an innovative self-evolving memory system designed for coding agents, organizing knowledge into a dynamic Zettelkasten-style graph. It allows memories to evolve and connect over time, enhancing an agent's ability to recall and utilize information effectively. This system offers both semantic and structural search capabilities for a richer knowledge base.
gh-aw-firewall: Secure Your Agentic Workflows with a Network Firewall
gh-aw-firewall is a robust network firewall designed specifically for agentic workflows, restricting outbound HTTP/HTTPS traffic to an allowlist of domains. It operates by running commands within a Docker sandbox, leveraging a Squid proxy for traffic filtering and an optional API proxy sidecar to securely manage LLM API keys. This project is a crucial component of GitHub's ongoing exploration into Agentic Workflows, enhancing their security and control.
mcp-gateway: Route MCP Servers and REST APIs Through One Endpoint
A Rust gateway that brings MCP servers and REST APIs behind one endpoint, with tools discovered on demand through a compact meta-tool surface. It also bridges MCP protocol revisions and adds centralized routing and policy controls.
code-session-memory: Automatic Vector Memory for AI Coding Sessions
code-session-memory provides automatic vector memory for various AI coding tools like OpenCode, Claude Code, Cursor, VS Code, Codex, and Gemini CLI. It indexes new messages into a vector database after each AI agent turn, enabling semantic search across all your past coding sessions. This tool ensures memory is shared across different platforms, enhancing developer productivity.
jan: Run AI Models Locally on Your Desktop
Jan is a cross-platform desktop app for running language models locally, with optional connections to cloud AI providers. It suits people who want a private local assistant, custom workflows, or an OpenAI-compatible local API.
TurboLLM: Run Local Language Models With GPU Tuning
TurboLLM runs local language models through a browser interface and OpenAI- or Anthropic-compatible APIs. It suits developers who want hardware-aware tuning and a choice of inference engines, including community forks.
mercury-agent: Run a Permission-Aware AI Agent Across Channels
Mercury is a self-hosted AI agent for CLI, Telegram, and web use, with permission prompts, persistent memory, token budgets, and scheduled tasks. It suits people who want an always-on assistant they can configure and operate locally.
awesome-free-models: Find Free AI Models, APIs, and Tools
A curated directory of free AI models, API tiers, and tools for local inference and AI development. Use it to compare options for experimentation or self-hosting, then verify current pricing and access terms with each provider.
SkillOpt: Train Reusable Skills for Frozen LLM Agents
SkillOpt improves natural-language agent skills using scored task trajectories and validation-gated edits, without changing the target model's weights. It is aimed at teams that can evaluate repeatable tasks and want to deploy a reusable skill document.
pi: Build and Customize AI Coding Agents
Pi is a TypeScript toolkit for building and using AI agents, with a coding-agent CLI, terminal UI, and unified API for multiple LLM providers. Its extensible design suits developers who want to adapt an agent to their workflows.
clowder-ai: Coordinate AI Agents as a Collaborative Team
Clowder AI adds a platform layer for coordinating agent CLIs from multiple model families. It is for developers who want agents to retain roles and shared memory, communicate with one another, and review each other’s work.