Archestra

Open-source AI infrastructure for controlling agent access, tools, and model spend

Linked from archestra.ai · checked 29 Oct

Tool type: Self-hosted AI control infrastructure
Made for: Platform, security, and engineering teams
Runs on: Self-hosted on Docker or Kubernetes; Runs locally via Tilt or Docker, or deploys to a Kubernetes cluster with Helm; the Docker quickstart requires at least 6 GB of memory.
Last checked: 29 Oct 2026; 2 product sources

What it does

Archestra is an open-source platform that sits between AI clients or agents and a company's systems. It takes in requests from clients such as Claude, Codex, Cursor, n8n, and internal agents, and routes them through an LLM proxy, MCP gateway, identity layer, policy engine, and audit trail before they reach models, MCP servers, knowledge sources, or business systems. The output is governed model and tool calls, each tied to an identity and logged.

Platform teams use it to roll out AI across a company without losing control of data, access, and actions. The distinctive part is that it covers the surrounding layers, not only one piece: a registry of MCP servers, a gateway, an agent runtime, chat for non-technical teams, cited knowledge retrieval from connected sources, and per-person, per-team, or per-agent spend budgets using virtual keys instead of shared API keys.

It suits platform and security teams running internal agents or MCP deployments. It is licensed under AGPL-3.0 and can be self-hosted through a local Docker or Tilt setup or deployed to Kubernetes with Helm. The FAQ states that teams under 30 users receive the enterprise license, including SSO, RBAC, and advanced knowledge bases, at no cost. Several features, such as Agent Runtime and Guardrails, are labeled Beta in the docs.

What it can do

LLM proxy that routes requests across model providers with server-side API keys, spend limits, and request tracing
MCP gateway and registry for connecting MCP servers with identity and audit controls
Identity-linked tool calls using Entra ID, Okta, OIDC, or SAML, with runtime credential resolution
Policy enforcement over tool access, approvals, isolation, and network egress before execution
Knowledge (RAG) answers with citations from connected enterprise sources and automatic source permission sync
Agents with instructions, skills, subagents, and triggers, runnable in Chat, Slack, Microsoft Teams, or via A2A

Works with

OpenAI
Anthropic
Azure OpenAI
Vertex AI
Bedrock
Kubernetes
Helm
Postgres
OpenTelemetry
Prometheus

Before you choose

Several features, including Agent Runtime, Guardrails, Hooks, and Model Pools, are labeled Beta in the docs, and Auto Routing is labeled Alpha.
Self-hosting means the team operates the infrastructure, databases, and model provider keys.
Enterprise features such as SSO, RBAC, and white-labeling require an enterprise license, which the FAQ states is free for teams under 30 users.

Listed under

Get started

Pull archestra/platform:latest and run the container, then sign in at localhost:3000, or follow the Get Started guide.

Install Docker, then run the quickstart image with the documented ports and volumes, and open http://localhost:3000.
For local development, clone the GitHub repository, run tilt up from the platform directory, and open localhost:3000.
For production, deploy with Helm and kubectl to a Kubernetes cluster, then connect identity providers and model providers.

Run the Docker quickstart

Plans and prices

Prices in USD

Self-hosting is free under AGPL-3.0, and teams under 30 users receive the enterprise license at no charge. Enterprise pricing above 30 users is not published; confirm the cost with Archestra. Model provider and infrastructure costs are separate.

Open source (self-hosted)FreeFree to self-host per FAQ.

Chat, agents, MCP gateway, LLM proxy, orchestrator, observability
AGPL-3.0 licensed

Enterprise (under 30 users)FreeFree for teams under 30 users per FAQ.

SSO and OIDC, RBAC, white-labeling, advanced knowledge bases with access control

Source: archestra.ai, checked 29 Oct 2026. Prices can change; check before you buy.

Sources for this overview

Product pages, documentation, and pricing checked 29 Oct 2026. Each source supports part of the overview; availability and prices can change.

Open-source infrastructure for enterprise AI | Archestra ↗archestra.ai
Archestra Docs | Archestra ↗archestra.ai

Ratings and reviews

No ratings for Archestra yet.

Be the first to rate it

What’s changed

Free account

See what’s changed at Archestra

Launches, price moves and pivots, each with a date and a source. Free with an account.

Signals

Free account

See Archestra’s signals

Launch traction, hiring and release activity, tracked over time. Free with an account.

Data and security

Free account

See how Archestra handles your data

Training on your files, compliance, hosting and the models behind it. Free with an account.

Compare with similar tools

Free account

Compare Archestra side by side

Prices, platforms and features against similar tools, with sources. Free with an account.

Where this comes from

How we research

Company website: archestra.ai: product, pricing and policy pages
YC directory mirror: Batch and company name
Web search: Confirms the official site
Social accounts: Accounts the company links, and its posts on X
Human review: Checked by a person
knowledge_base
manual_review

Last checked 29 Oct 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.