---
title: "Agent Memory Enterprise"
description: "The SurrealDB Agent Memory Enterprise plan: single-tenant deployments, your own language models, an agreed token volume and support with SLAs, with self-hosted deployments in development."
canonical_url: "https://surrealdb.com/docs/agent-memory/operate/enterprise"
section: "Agent Memory"
surrealdb_version: "4.x"
last_modified: "2026-11-18T02:09:16Z"
---

> Full SurrealDB documentation index: https://surrealdb.com/docs/llms.txt

# Agent Memory Enterprise

The SurrealDB Agent Memory Enterprise plan: single-tenant deployments, your own language models, an agreed token volume and support with SLAs, with self-hosted deployments in development.

Agent Memory Enterprise is the SurrealDB Agent Memory plan for organisations that need a deployment of their own. It has the same memory features as the other plans. It adds control over where Agent Memory runs, which language models it calls and how many tokens it can use.

> [!NOTE]
> Single-tenant deployments are available on the Enterprise plan. Self-hosted deployments, which run in your own cloud account or data centre, are in development as part of the same plan. To discuss either, [contact sales](https://surrealdb.com/contact/sales).

## What Enterprise adds

| Feature | Sandbox, Lite, Standard and Plus | Enterprise |
| --- | --- | --- |
| Deployment | Contexts on a shared SurrealDB Cloud deployment | A single-tenant deployment. Self-hosted deployments in your own infrastructure are in development |
| Language models | Google, OpenAI, Anthropic, Cerebras and OpenRouter models, chosen per Context | The same, plus your own model through an OpenAI-compatible endpoint |
| Air-gapped deployment | Not available | In development, as a form of self-hosted deployment |
| Tokens | The plan's token allowance | A volume agreed with SurrealDB |
| Support | Community support | Enterprise support with SLAs |

## What stays the same

Every plan has the same memory features: the knowledge graph with entity extraction, conversational memory, tri-temporal versioning and supersession, provenance on every fact, hybrid retrieval, traces, and the multimodal document pipeline. Every plan also has the same isolation and security features: one database per Context, scoped permissions with attenuated API keys, prompt-injection scanning, and the [forget](/docs/agent-memory/learn/operations/forget.md) operations.

The REST API, the MCP server and the SDKs are also the same. Code written against a Context on SurrealDB Cloud works against an Enterprise deployment after you change the endpoint, the Context ID and the API key.

## Deployment options

- **Single-tenant:** SurrealDB runs an Agent Memory deployment that serves only your organisation. SurrealDB applies the deployment settings agreed with you.
- **Self-hosted (in development):** Agent Memory runs in your own cloud account or data centre, in front of your own SurrealDB deployment and object storage, and you set the deployment settings yourself.
- **Air-gapped (in development):** a self-hosted deployment with no connection to the internet, arranged with SurrealDB as part of an Enterprise agreement.

### Why choose a deployment of your own

Data isolation is the same on every plan: each Context has its own database, whichever deployment it runs on. A single-tenant deployment adds control over the settings that apply to a whole deployment, which can change only where every Context belongs to one organisation:

- **The [default language model](/docs/agent-memory/operate/enterprise/configuration.md#language-models)**, and [your own model](/docs/agent-memory/operate/enterprise/configuration.md#your-own-model) through an OpenAI-compatible endpoint, so that your prompts and data go only to a model you have approved or host yourself.
- **[Storage](/docs/agent-memory/operate/enterprise/configuration.md#storage)**, for the database and object storage behind your Contexts.
- **[Limits](/docs/agent-memory/operate/enterprise/configuration.md#capacity-and-limits)**, sized for your workload rather than for a shared deployment.

A deployment that serves only your organisation also carries only your organisation's load. SurrealDB runs the deployment for you, so a single-tenant deployment gives you this control without operating it yourself.

## What you can configure

Settings apply at two levels. Per-Context settings, such as the model for each stage and the response cache, are available on every plan. Deployment settings, such as the default language model, your own model endpoint, storage and limits, apply to every Context on a deployment, so you can change them only on a deployment of your own. Some behaviour, such as when `context` facts expire, is the same on every deployment.

[Enterprise configuration](/docs/agent-memory/operate/enterprise/configuration.md) lists every setting at each level, and the behaviour that stays fixed.

## Get Enterprise

Enterprise pricing is agreed with SurrealDB. To discuss a deployment, [contact sales](https://surrealdb.com/contact/sales). To compare plans, see [Agent Memory pricing](https://surrealdb.com/pricing?product=agentMemory).

## Related reading

- [Enterprise configuration](/docs/agent-memory/operate/enterprise/configuration.md) - per-Context settings, deployment settings and fixed behaviour
- [Configuration reference](/docs/agent-memory/reference/configuration.md) - every per-Context field and its default
- [Models per stage](/docs/agent-memory/learn/tuning/models-per-stage.md) - which model runs at each stage
- [Agent Memory on SurrealDB Cloud](/docs/agent-memory/operate/cloud.md) - Contexts and API keys on the shared plans
