# Agno > Agno is a self-driving agent platform that runs in your cloud. Use coding agents to build, deploy and improve customer-facing and internal agents, teams and workflows in pure Python with the Agno SDK, run them in your own cloud on the AgentOS runtime, serve them in your product over REST and in ChatGPT and Claude over MCP, and manage them from the Agno Control Plane. The SDK and AgentOS are open source. Agno is model-agnostic: bring OpenAI, Anthropic, Google or any supported provider and switch without rewriting your agents. The full developer documentation lives at https://docs.agno.com (LLM index: https://docs.agno.com/llms.txt). ## Products - [AgentOS](https://www.agno.com/products/runtime): Agent runtime — turns the agents you build into a production service with one command, in your cloud, on your database, behind your auth - [Agent SDK](https://www.agno.com/products/sdk): The Agent SDK for coding agents: three primitives (agents, teams and workflows) with capabilities you attach as your use case needs them - [Control Plane](https://www.agno.com/products/control-plane): Watch every agent, team and workflow you run and step in when it matters, from one place ## Compare - [Benchmarks](https://www.agno.com/benchmarks): Agent instantiation time and memory for Agno, LangGraph and CrewAI, measured with a public harness. Hardware, versions and dates included - [Agno vs LangGraph](https://www.agno.com/compare/agno-vs-langgraph): Plain Python vs compiled graphs, and who ships the runtime. Every claim carries a source - [Agno vs CrewAI](https://www.agno.com/compare/agno-vs-crewai): Role-based crews for prototypes vs a runtime built for production serving - [Agno vs OpenAI Agents SDK](https://www.agno.com/compare/agno-vs-openai-agents-sdk): A minimal single-platform SDK vs a model-agnostic platform you own ## Pricing - [Pricing](https://www.agno.com/pricing): The runtime is free at any scale, including serving. Pro is $150/month for the live control plane; Enterprise is custom - [Enterprise](https://www.agno.com/enterprise): Self-hosted control plane, SAML SSO, custom RBAC, audit logging and dedicated support ## Articles - [Agno Articles](https://www.agno.com/articles): Engineering AI agents at scale — the 10 newest articles below; every article is in llms-full.txt - [Community Roundup: September 2026](https://www.agno.com/articles/community-roundup-september-2026.md): Agno's September recap: PublicSurface puts agents in front of anonymous users, page storage turns your docs into agent knowledge, rerankers work on every vector database, and AgentOS MCP servers scale across replicas. - [Agent evals are the new unit tests (and most teams skip them)](https://www.agno.com/articles/agent-evals-are-the-new-unit-tests.md): Agent evals verify what an AI agent actually did. Why prompt changes cause regressions, and how to build a minimum viable eval loop in CI. - [Best frameworks for building customer-facing AI agents in 2026](https://www.agno.com/articles/best-frameworks-for-customer-facing-ai-agents.md): Compare the best AI agent frameworks for customer-facing agents in 2026 on per-user isolation, auth, approvals, and embedding. - [A pile of MCP tools is not an agent](https://www.agno.com/articles/mcp-server-vs-agent.md): An MCP server tells an assistant what your product can do. It can't tell it what to do. When a flat tool list breaks, and when it's enough. - [Community Roundup: August 2026](https://www.agno.com/articles/community-roundup-august-2026.md): Agno's August recap: v3.0 ships with durable background execution, a dedicated runs table, Studio governance and platform-wide user isolation — and an agent scores 100% on the ARC-AGI-3 public set. - [How to build an agent for your product and serve it everywhere](https://www.agno.com/articles/how-to-build-an-agent-for-your-product.md): A guide for engineering leaders building an agent for their product. One agent, served in your product, in Claude and ChatGPT, and in Slack. - [Learning Machines: the memory that cracked ARC-AGI-3 public set](https://www.agno.com/articles/learning-machines.md): What happens when we let agents learn. - [With v3.0, we went after what actually breaks agents in production](https://www.agno.com/articles/what-actually-breaks-agents-in-production.md): Agno v3.0 hardens AI agents for production workloads: a dedicated runs table, tool result and media offloading, durable background execution, per-user isolation, Studio governance, and CodeMode. Includes the required migration. - [Agno 3.0 and the agent platform that builds itself](https://www.agno.com/articles/agno-3-0.md): We're excited to release Agno 3.0, bringing us one step closer to our vision of an agent platform that builds itself. - [100% on the ARC-AGI-3 public set](https://www.agno.com/articles/arc-agi-arcade.md): An agent powered by GPT-5.6 scored 100.00 RHAE on the ARC-AGI-3 public set: 183 levels across 25 games, verified by an official ARC scorecard. - [Agno Changelog](https://www.agno.com/changelog.md): Every shipped feature, fix and improvement, grouped by the day it shipped ## Docs - [Agno Documentation](https://docs.agno.com): Developer docs for the SDK and AgentOS — guides, examples, API reference - [Docs llms.txt](https://docs.agno.com/llms.txt): The documentation's own LLM index — the counterpart to this file ## Optional - [Customers](https://www.agno.com/customers): Teams running production multi-agent systems on Agno - [About](https://www.agno.com/about): Who builds Agno and where to find us - [Community](https://www.agno.com/community): The Agno developer community — Discord, GitHub, events - [Changelog](https://www.agno.com/changelog): Every feature, fix and improvement we ship, by the day it shipped - [Full site content for LLMs](https://www.agno.com/llms-full.txt): This index plus full page copy and every blog post as markdown