Open source at the core. Simple plans in the cloud.
Run it yourself for free, start on the hosted Free plan, bring your own model key for more, or talk to us about an enterprise deployment.
- 100% MIT Licensed codebase
- Run locally with Docker, Compose, or Kubernetes
- Postgres-journaled event log & crash recovery
- Zero usage limits — it's your own hardware
- Connect your own LLM keys (OpenAI, Anthropic, Gemini, Ollama)
- Full access to tools, memory, and code interpreter
Explorer (BYOK)
Connect your own OpenAI, Anthropic, Gemini or OpenRouter key.
- 100 messages a day, billed to your own provider account
- No markup on model costs: calls go straight to your provider
- Scheduled background agents and recurring tasks
- Persistent history, documents and memory
- The flat fee covers our processing: parsing, embeddings, storage
- Custom multi-agent architecture design & evals
- Private cloud / on-premises VPC deployment
- Bespoke tool building & proprietary API integration
- Direct Slack channel with core platform engineers
- Enterprise SLA, SSO, and compliance support
- Dedicated hands-on onboarding & code reviews
Frequently asked questions
Everything you need to know about licensing, limits and enterprise deployment.
How does the Open Source version differ from the hosted platform?
The underlying engine is 100% open source under the MIT license. You can clone the repository and run Agent Substrate on your laptop or Kubernetes cluster with no limits. The hosted platform simply runs the infrastructure (databases, sandboxes, worker pools) for you so you can focus on building agents without managing servers.
What happens when I reach my daily limit?
You'll see a clear message with the time your limit resets, and the usage meter in the sidebar always shows what is left. Nothing is charged automatically. To send more, connect your own API key.
When do I start paying for Explorer?
Billing is not switched on yet. Explorer limits apply today and the monthly fee will be collected once checkout opens; you'll be told before anything is charged.
Do you mark up model API costs?
No. When you use your own API keys (BYOK), calls go directly between your provider account and the agent runtime. We never take a percentage or markup on token consumption.
What does the Consulting & Enterprise engagement include?
Our core engineering team works directly with your company to architect bespoke agent pipelines, build custom tools and private database connectors, deploy the substrate inside your private VPC, and optimize agent reliability and evaluation benchmarks.
Can I upgrade from self-hosting to hosted, or vice-versa?
Yes. Because both use the same underlying Agent Substrate engine and PostgreSQL schema, workflows and agent definitions are fully portable between your self-hosted instance and our cloud.