Eliza
Open source agentic operating system
Last verified:
What is Eliza?
elizaOS is an open-source, TypeScript-based agentic operating system for building, orchestrating, and deploying autonomous AI agents that can think, learn, and take actions across interfaces and services. It provides a lightweight CLI and framework that gets a working agent online in minutes while supporting production scale deployments and persistent memory so agents retain context across conversations. The platform is plugin-driven and modular, enabling connectors for messaging platforms (Discord, Telegram, X), model providers (OpenAI, Llama, Anthropic, etc.), blockchains, databases, and custom actions so teams can hot-swap components without touching core logic. ElizaOS targets developers, teams, and organizations who want to build specialized autonomous agents (customer support, trading bots, research assistants, NPCs, smart-home controllers, growth automation, DAO delegates, etc.) while offering options to deploy locally, via Docker, to Eliza Cloud, or into existing infrastructure.
Eliza pricing
Pricing model: Freemium
The website describes both local/self-hosted usage of the open-source framework (free) and a managed Eliza Cloud offering; local use is free as an open-source project while Eliza Cloud provides a hosted, no-code agent creator, API keys, and cloud deployment capabilities (tiered paid plans for managed hosting and commercial support are offered via Eliza Cloud). Exact plan names, monthly rates, quotas, and feature-by-feature inclusions for paid tiers are presented in Eliza Cloud’s dashboard and documentation rather than on the main marketing pages.
Eliza pros
- Open-source TypeScript framework suitable for developer customization
- Three-command quickstart (install CLI, create, start) for fast iteration
- Plugin registry with 90+ plugins including social, blockchain, and provider adapters
- Persistent memory so agents remember and learn from interactions
- Unified message bus supporting Discord, Telegram, X, HTTP and on-chain
- Composable multi-agent architecture (Worlds and Rooms) for delegation
- Strategic action chaining for branching, parallel steps, scheduling and pausing
- Model-agnostic support (OpenAI, Llama, Grok, Anthropic and more)
- Deploy anywhere: local, Docker, Eliza Cloud, or custom infra
- CLI and REST API for both no-code and programmatic workflows
- Built-in agent creator/no-code editor available in cloud offering
- Extensible character/personality files to define agent behavior and tone
- Hot-swappable modules to avoid vendor lock-in and keep the core lightweight
- Production-ready design that scales from single agents to millions of interactions
- Active community and many contributors with extensive documentation and examples
Eliza cons
- Developer-focused: requires familiarity with TypeScript/CLI for advanced uses
- Requires configuration for databases and model provider keys for full features
- On-prem or custom deployments need infrastructure knowledge (Docker, hosting)
- Plugin quality and maintenance may vary between community contributors
- Running large-scale agents can incur significant model provider costs
- Some features rely on third-party provider APIs (vendor dependencies)
- Eliza Cloud advanced features may be behind paid tiers (cloud-specific)
- Learning curve for multi-agent orchestration concepts (Worlds, Rooms)
Frequently asked questions about Eliza
How do I create my first elizaOS agent?
Use the elizaOS CLI: install the CLI globally, run elizaos create to scaffold a project, configure your database and model provider during the interactive setup, then run elizaos start to launch the agent and open the local web chat at http://localhost:3000.
Which model providers and models does elizaOS support?
elizaOS is model-agnostic and supports providers such as OpenAI, Llama-family models, Anthropic, and community adapters; you select your provider and model in the project configuration or cloud agent settings and supply the necessary API key.
Can I deploy elizaOS agents to production?
Yes — agents can be deployed locally, in Docker containers, to Eliza Cloud, or to your own infrastructure; the framework is designed to scale from development to millions of interactions with persistent memory and production-oriented plugins.
What integrations and plugins are available?
There is a plugin registry with 90+ plugins covering messaging platforms (Discord, Telegram, X), blockchains (Ethereum, Solana), vector stores, databases, model providers, and many custom action connectors; plugins are distributed as npm packages you can hot-swap at runtime.
How does elizaOS handle memory and context?
Agents use persistent memory to remember and learn from every interaction; memory backends can be configured (local stores or external databases) so agents maintain long-term context across rooms, worlds, and deployments.
Do I need to code to use elizaOS Cloud?
No — Eliza Cloud provides a no-code Agent Creator for building and customizing agents through a visual interface, though programmatic options like the REST API and CLI remain available for developers who prefer code-first workflows.
How do agents communicate across channels and with each other?
elizaOS uses a unified message bus and the Worlds/Rooms architecture so agents keep localized context while emitting events across transports; this enables delegation, consensus, and load balancing between agents without rewriting business logic.
What are strategic action chains and when should I use them?
Strategic action chaining lets agents perform LLM-driven tool calls in sequences that can branch, run in parallel, schedule future steps, or pause for user input, making them suitable for complex workflows like trading, orchestration, or multi-step automation.
How do I add or build a custom plugin?
Follow the documentation guide to create a plugin package (npm) that implements the required plugin hooks; once published or installed locally you can enable it in your project to extend capabilities such as new transports, actions, or storage backends.
What support and community resources are available?
Comprehensive documentation, quickstarts, CLI and REST references, a plugin registry, GitHub repository, and an active contributor community provide guides, examples, and assistance for development and deployment tasks.