
# AI applications & RAG

AI application builders, chat interfaces, agents and retrieval-augmented generation.

Canonical page: https://launchinyourcloud.com/categories/ai-applications-rag/
Catalog revision: 170de95bb7d832897d269051ad4500713acf96a62f3d01bc1d4e649679a56fde



## Related questions

- [How do I run ChatGPT-style AI privately in my own cloud?](https://launchinyourcloud.com/questions/private-chat-ai/)
- [What self-hosted RAG and LLM app platforms are available?](https://launchinyourcloud.com/questions/rag-llm-platforms/)


Named clouds have provider-specific evidence. Missing clouds do not establish incompatibility. Preserve edition, account boundary, operating responsibilities and publisher disclosures. Source review dates record human checks, not rebuilds.


AI application builders, chat interfaces, agents and retrieval-augmented generation.

8 listings.


## AnythingLLM Docker

Self-hosted application for document-grounded AI chat, retrieval and agents.

- Listing: https://launchinyourcloud.com/directory/anythingllm/
- Publisher listing: No
- Entry kind: product
- Documented cloud IDs: aws, gcp, railway
- Operating model: Customer-operated
- Infrastructure IDs: servers
- License: MIT License for upstream AnythingLLM; model weights, connectors and provider services have separate terms.
- Licensing model: open-source
- Installation: Run the official Docker edition on a customer-operated host with persistent application storage and configured model/vector endpoints.
- Architecture: Customer-operated web application and document/retrieval storage with configurable inference providers
- Operations: Customer manages users, documents, persistent storage, model access, integration secrets, backups and upgrades.
- Product website: https://anythingllm.com/
- Source review: 2026-11-18; cloud review: 2026-11-12; licensing review: 2026-11-10; infrastructure review: 2026-11-10

### Scope and qualifications

The Docker edition supports multi-user operation and differs from the desktop application. Local inference is an option; configured external LLMs, embeddings, vector stores and agent tools can send data outside the account. AnythingLLM v1.20.0 hardens embed history, extension API-key authorization, workspace-agent access, password recovery, framing, and document-tool downloads. Optional MCP identity forwarding creates a new outbound identity path; keep the server and extension on compatible 1.20+ versions.

MIT License for upstream AnythingLLM; model weights, connectors and provider services have separate terms. This classification covers the reviewed upstream edition; external services and dependencies retain their own terms.

Customer-operated web application and document/retrieval storage with configurable inference providers. This is a customer-operated server/container deployment; portability does not establish named cloud-provider support.


### Category fit

- ai-applications-rag: Multi-user application for document-grounded chat, RAG and agents with configurable model and vector-database integrations.

- ai-tools: Self-hosted AI application interface and retrieval/agent integrations; model inference is separately configured.

### Deployment boundaries


- aws: Customer cloud account; Customer-operated; control plane: not-assessed. Source: https://docs.anythingllm.com/installation-docker/overview (checked 2026-11-12). 

- gcp: Customer cloud account; Customer-operated; control plane: not-assessed. Source: https://docs.anythingllm.com/installation-docker/overview (checked 2026-11-12). 

- railway: Customer platform workspace; Customer-operated; control plane: not-assessed. Source: https://docs.anythingllm.com/installation-docker/overview (checked 2026-11-12). 


### Cloud deployment evidence

- AnythingLLM Docker installation overview (aws, gcp, railway): https://docs.anythingllm.com/installation-docker/overview
  The project explicitly names AWS, Google Cloud and Railway for self-hosting the Docker edition, paired with its Docker installation workflow. This is an upstream hosting compatibility statement. Model endpoints, embeddings and external vector stores are separately configured and can have different data boundaries.
  Source type: vendor-documentation; reviewed: 2026-11-12; vendor support: not-assessed.


### Community and maintenance

Public release notes credit named contributors and new contributors across multiple releases, alongside an established open contribution process.

Reviewed 2026-11-10.

- Release maintenance and contributor credits: https://github.com/Mintplex-Labs/anything-llm/releases

- Community contribution process: https://github.com/Mintplex-Labs/anything-llm

### Sources

- Official installation and runtime: https://docs.anythingllm.com/installation-docker/overview

- Upstream project and capabilities: https://github.com/Mintplex-Labs/anything-llm

- Current upstream license: https://raw.githubusercontent.com/Mintplex-Labs/anything-llm/master/LICENSE

- AnythingLLM v1.20.0 release: https://github.com/Mintplex-Labs/anything-llm/releases/tag/v1.20.0

### Licensing sources

- https://raw.githubusercontent.com/Mintplex-Labs/anything-llm/master/LICENSE

### Infrastructure sources

- https://docs.anythingllm.com/installation-docker/overview


### Buyer diligence


Research coverage: 6/12 fields (50%); 6 documented, 0 unresolved. Coverage is not a feature score.

Unrecorded fields are not researched, not a statement that the feature is unavailable. Sources checked — unresolved means the cited material did not establish an answer. Hosted-service certifications do not automatically cover customer installations.


- Marketplace delivery: Not researched.

- Cloud commitment eligibility: Not researched.

- Single sign-on: Documented. The self-hosted Docker configuration offers Simple SSO passthrough: a trusted application uses an API key to issue short-lived single-use login links for existing users. It requires multi-user mode and a configured admin. This is a bespoke integration, not native SAML/OIDC federation; the vendor recommends internally exposed instances. Scope: Docker Simple SSO passthrough; not standards-based federation. Source: https://docs.anythingllm.com/configuration (reviewed 2026-11-20).

- SCIM provisioning: Not researched.

- Compliance attestations: Not researched.

- Security controls: Documented. Current advisory review: AnythingLLM v1.20.0 hardens embed history, extension API-key authorization, workspace-agent access, password recovery, framing, and document-tool downloads. Upgrade the server and extension to compatible 1.20+ releases and review optional MCP identity forwarding. Scope: Upstream v1.20.0 release notes; no consolidated affected-version range is asserted.. Source: https://github.com/Mintplex-Labs/anything-llm/releases/tag/v1.20.0 (reviewed 2026-11-18).

- Air-gapped operation: Not researched.

- Deployment artifact / IaC: Documented. The Docker guide documents a prebuilt application image or source-built Compose setup. Persistent storage and user permissions must be configured; local model/vector endpoints and their resources are separate. Scope: Upstream self-hosted deployment; not a provider support or production-readiness certification. Source: https://docs.anythingllm.com/installation-docker/local-docker (reviewed 2026-11-14).

- Minimum footprint / sizing: Documented. The Docker requirements list 2 GB RAM, a two-core CPU with AVX2 and 5 GB storage as a small starting profile. Model inference, embeddings, vector storage, documents and concurrency can require much more; verify AVX2 passthrough on virtual machines. This is not an HA or 1-TB production recommendation. Scope: Self-hosted Docker baseline; model services, data and HA additional. Source: https://docs.anythingllm.com/installation-docker/system-requirements (reviewed 2026-11-20).

- License pricing: Documented. The reviewed upstream core is MIT licensed. Optional services, bundled dependencies, commercial distributions and support can have separate terms. This records the source licence, not an infrastructure or operating-cost estimate. Model weights and external model APIs require their own review. Scope: Upstream core licence; commercial features, support and hosting priced separately. Source: https://raw.githubusercontent.com/Mintplex-Labs/anything-llm/master/LICENSE (reviewed 2026-11-20).

- Operating cost estimate: Not researched.

- Release snapshot / cadence: Documented. AnythingLLM v1.20.0 was published November 17, 2026. It includes access and session hardening plus MCP identity forwarding and timeout changes; keep server and extension versions compatible. Scope: Upstream AnythingLLM self-hosted release. Source: https://github.com/Mintplex-Labs/anything-llm/releases/tag/v1.20.0 (reviewed 2026-11-18).




### Security and lifecycle reviews

Selected editorial reviews only; no flag does not mean safe or supported.

- AnythingLLM 1.20.0 access hardening (security-advisory): AnythingLLM v1.20.0 hardens embed history, extension API-key authorization, workspace-agent access, password recovery, framing, and document-tool downloads. Scope: Upstream v1.20.0 release notes; no consolidated affected-version range is asserted. Conditions: Applicability depends on enabled embedding, extensions, agents, recovery, and document tools. Action: Upgrade the server and extension to compatible 1.20+ releases and review optional MCP identity forwarding. Source: https://github.com/Mintplex-Labs/anything-llm/releases/tag/v1.20.0; reviewed 2026-11-18.




## Dify Self-Hosted

Visual AI application and RAG workflow platform available as a self-hosted container stack.

- Listing: https://launchinyourcloud.com/directory/dify/
- Publisher listing: No
- Entry kind: product
- Documented cloud IDs: alibaba
- Operating model: Customer-operated
- Infrastructure IDs: servers
- License: Modified Apache 2.0 terms; multi-workspace service operation and frontend branding can require commercial authorization.
- Licensing model: source-available
- Installation: Official Docker Compose deployment; pin a release and configure its required services.
- Architecture: Customer-operated application, workers, PostgreSQL and vector-store services
- Operations: Customer operates the containers, persistent stores, secrets, backups and upgrades; commercial edition support varies.
- Product website: https://dify.ai/
- Source review: 2026-11-18; cloud review: 2026-11-12; licensing review: 2026-11-10; infrastructure review: 2026-11-10

### Scope and qualifications

Self-hosting Dify does not itself host every model used by a flow. Model-provider plugins, embedding services, tools and connectors may send inputs or retrieved content outside the deployment. Select local endpoints where required and review each dependency. This entry does not conflate the generic self-hosted stack with the separately licensed AWS Premium edition. Dify 1.20.0 changes deployment and API behavior: OPENAPI_ENABLED defaults to true, shared.env is retired, six migrations apply from 1.19, and HITL, plugin-daemon, OpenShell, and optional file-scanning requirements change. It also fixes OAuth callback reuse, member-role enforcement, vector deletion scope, public stream detail exposure, and CVE-2026-116591.

Classified here as source-available because the upstream license modifies Apache 2.0 with multi-workspace service and frontend-branding conditions. Some uses require separate commercial authorization; this is not an unmodified Apache-licensed core.

Docker Compose runs application services, workers and persistent database/vector-store containers on a host. This is provisioned server/container infrastructure; using a managed external model does not make this platform serverless.


### Category fit

- ai-applications-rag: Visual application, agent-workflow and retrieval-augmented generation platform. The application platform is deployable; selected models and plugins are separate dependencies.

- ai-tools: Visual AI application and RAG workflow platform available as a self-hosted container stack.

### Deployment boundaries


- alibaba: Customer cloud account; Customer-operated; control plane: not-assessed. Source: https://www.alibabacloud.com/help/en/openapi/integrating-openapi-mcp-server-in-dify (checked 2026-11-12). Source: https://help.aliyun.com/en/compute-nest/use-cases/dify-community-edition-service-instance-deployment-document (checked 2026-11-12). 


### Cloud deployment evidence

- Dify Docker deployment on Alibaba Cloud ECS (alibaba): https://www.alibabacloud.com/help/en/openapi/integrating-openapi-mcp-server-in-dify
  Alibaba Cloud documents cloning upstream Dify and running its Docker deployment on a customer ECS instance. The example then integrates an MCP server; model providers and other outbound integrations must be reviewed separately.
  Source type: provider-guide; reviewed: 2026-11-12; vendor support: not-assessed.

- Dify Community Edition full application on Alibaba Compute Nest (alibaba): https://help.aliyun.com/en/compute-nest/use-cases/dify-community-edition-service-instance-deployment-document
  Alibaba documents a Dify Community Edition service instance that creates ECS compute in the customer-selected VPC, zone and region, then exposes the installed Dify application for administrator registration. This is application hosting rather than an MCP/model connector, and is separate from the licensed Dify AWS Premium offering. Review model endpoints and release compatibility.
  Source type: provider-guide; reviewed: 2026-11-12; vendor support: not-assessed.


### Sources

- Docker Compose deployment: https://docs.dify.ai/en/self-host/deploy/quick-start/docker-compose

- Application capabilities and integrations: https://github.com/langgenius/dify

- Current license and additional conditions: https://github.com/langgenius/dify/blob/main/LICENSE

- Dify Community Edition full application on Alibaba Compute Nest: https://help.aliyun.com/en/compute-nest/use-cases/dify-community-edition-service-instance-deployment-document

- Dify 1.20.0 release: https://github.com/langgenius/dify/releases/tag/1.20.0

### Licensing sources

- https://github.com/langgenius/dify/blob/main/LICENSE

### Infrastructure sources

- https://docs.dify.ai/en/self-host/deploy/quick-start/docker-compose


### Buyer diligence


Research coverage: 6/12 fields (50%); 6 documented, 0 unresolved. Coverage is not a feature score.

Unrecorded fields are not researched, not a statement that the feature is unavailable. Sources checked — unresolved means the cited material did not establish an answer. Hosted-service certifications do not automatically cover customer installations.


- Marketplace delivery: Not researched.

- Cloud commitment eligibility: Not researched.

- Single sign-on: Edition-dependent. SAML, OIDC and OAuth2 SSO appear in the self-hosted Enterprise feature comparison and are enabled by an Enterprise licence. This does not establish those features for an unlicensed Community installation; confirm identity-provider and end-user/admin coverage. Scope: Self-hosted Enterprise licence; Community differs. Source: https://dify.ai/pricing/dify-enterprise (reviewed 2026-11-20).

- SCIM provisioning: Not researched.

- Compliance attestations: Not researched.

- Security controls: Documented. Current advisory review: Dify 1.20.0 fixes OAuth callback reuse, member-role enforcement, vector deletion scope, public stream detail exposure, CVE-2026-116591, and security dependencies. Apply the 1.20.0 migrations and configuration changes, or a later supported release, and retest authentication and public endpoints. Scope: Upstream Dify 1.20.0 release; consult its linked advisories for component-specific ranges.. Source: https://github.com/langgenius/dify/releases/tag/1.20.0 (reviewed 2026-11-18).

- Air-gapped operation: Not researched.

- Deployment artifact / IaC: Documented. The upstream Compose file is generated from its template and environment configuration. Use a consistent release checkout rather than mixing generated files or dependency settings across releases. Scope: Upstream multi-service deployment; models and integrations need separate configuration. Source: https://github.com/langgenius/dify/blob/main/docker/docker-compose.yaml (reviewed 2026-11-14). Deployment artifact: https://github.com/langgenius/dify/blob/main/docker/docker-compose.yaml. Artifact reference: main (moving branch).

- Minimum footprint / sizing: Documented. The Compose quickstart specifies at least 2 CPU cores and 4 GiB RAM. The selected database, vector store, models and workload may require more resources. Scope: Quickstart application stack; not model-serving or production HA sizing. Source: https://docs.dify.ai/en/self-host/deploy/quick-start/docker-compose (reviewed 2026-11-14).

- License pricing: Documented. Dify uses modified Apache-2.0 terms, with additional conditions for multi-tenant workspaces and frontend branding. Certain uses require a commercial licence. Do not treat it as unmodified Apache-2.0; Enterprise, support and model-provider charges are separate. Scope: Upstream core licence; commercial features, support and hosting priced separately. Source: https://raw.githubusercontent.com/langgenius/dify/main/LICENSE (reviewed 2026-11-20).

- Operating cost estimate: Not researched.

- Release snapshot / cadence: Documented. Dify 1.20.0 was published November 18, 2026. It includes six migrations from 1.19, retires shared.env, changes OPENAPI_ENABLED and HITL behavior, and requires review of plugin-daemon, OpenShell, and file-scanning configuration. Scope: Upstream Dify self-hosted 1.20 release. Source: https://github.com/langgenius/dify/releases/tag/1.20.0 (reviewed 2026-11-18).




### Security and lifecycle reviews

Selected editorial reviews only; no flag does not mean safe or supported.

- Dify 1.20.0 security update (security-advisory): Dify 1.20.0 fixes OAuth callback reuse, member-role enforcement, vector deletion scope, public stream detail exposure, CVE-2026-116591, and security dependencies. Scope: Upstream Dify 1.20.0 release; consult its linked advisories for component-specific ranges. Conditions: Exposure depends on enabled OAuth, public application streams, role configuration, vector stores, and affected dependencies. Action: Apply the 1.20.0 migrations and configuration changes, or a later supported release, and retest authentication and public endpoints. Source: https://github.com/langgenius/dify/releases/tag/1.20.0; reviewed 2026-11-18.




## Langflow Self-Hosted

Visual AI workflow and agent builder with Docker and persistent PostgreSQL deployment.

- Listing: https://launchinyourcloud.com/directory/langflow/
- Publisher listing: No
- Entry kind: product
- Documented cloud IDs: aws, azure, gcp
- Operating model: Customer-operated
- Infrastructure IDs: servers
- License: MIT for the upstream project; connected models, providers and bundles have their own terms.
- Licensing model: open-source
- Installation: Official Docker image or Docker Compose; Python installation and Kubernetes options are also documented.
- Architecture: Customer-operated flow application with persistent database and configured model/tool endpoints
- Operations: Customer manages the application, database, access controls, installed components, backups and version upgrades.
- Product website: https://www.langflow.org/
- Source review: 2026-11-18; cloud review: 2026-11-12; licensing review: 2026-11-10; infrastructure review: 2026-11-10

### Scope and qualifications

The reviewed model-component docs support Ollama and configurable OpenAI-compatible endpoints as well as commercial providers. External models, embeddings, tools or custom components can move data beyond the host. Hosting Langflow is distinct from hosting those dependencies and from any vendor-managed service. Langflow 1.13.3 enforces custom-component policy in background workflows, invalidates native sessions after password changes, updates security dependencies, and aligns Docker and wheel packaging. Recheck custom components and session behavior after upgrading.

The reviewed upstream Langflow project uses the MIT license. This classification covers that project; connected model weights, provider services and third-party components have their own terms.

Official images run the application on a Docker host; the Compose deployment adds persistent PostgreSQL storage. Model-serving hardware is a separate requirement if inference is also self-hosted.


### Category fit

- ai-applications-rag: Visual builder and runtime for AI flows, agents and model integrations. Local model endpoints are configurable; a flow can also call externally hosted providers.

- ai-tools: Visual AI workflow and agent builder with Docker and persistent PostgreSQL deployment.

### Deployment boundaries


- aws: Customer cloud account; Customer-operated; control plane: not-assessed. Source: https://docs.langflow.org/deployment-architecture (checked 2026-11-12). 

- azure: Customer cloud account; Customer-operated; control plane: not-assessed. Source: https://docs.langflow.org/deployment-architecture (checked 2026-11-12). 

- gcp: Customer cloud account; Customer-operated; control plane: not-assessed. Source: https://docs.langflow.org/deployment-architecture (checked 2026-11-12). 


### Cloud deployment evidence

- Langflow Kubernetes deployment architecture (aws, azure, gcp): https://docs.langflow.org/deployment-architecture
  The project explicitly identifies EKS, AKS and GKE for its Helm deployment. Separate the development IDE from the production runtime, and review PostgreSQL, model endpoints and external vector stores.
  Source type: vendor-documentation; reviewed: 2026-11-12; vendor support: not-assessed.


### Sources

- Docker and Docker Compose deployment: https://docs.langflow.org/deployment-docker

- Model components and local endpoints: https://docs.langflow.org/components-models

- Upstream MIT license: https://github.com/langflow-ai/langflow/blob/main/LICENSE

- Langflow 1.13.3 release: https://github.com/langflow-ai/langflow/releases/tag/v1.13.3

### Licensing sources

- https://github.com/langflow-ai/langflow/blob/main/LICENSE

### Infrastructure sources

- https://docs.langflow.org/deployment-docker


### Buyer diligence


Research coverage: 6/12 fields (50%); 6 documented, 0 unresolved. Coverage is not a feature score.

Unrecorded fields are not researched, not a statement that the feature is unavailable. Sources checked — unresolved means the cited material did not establish an answer. Hosted-service certifications do not automatically cover customer installations.


- Marketplace delivery: Not researched.

- Cloud commitment eligibility: Not researched.

- Single sign-on: Documented. Current 1.13 documentation supports an upstream OIDC proxy/SSO gateway forwarding tokens that Langflow validates using issuer, audience and JWKS. It can provision a local user; built-in credentials remain a separate path. Configure the gateway and validation explicitly and replace development identity-provider examples. Scope: Langflow 1.13 external-token validation with a separately operated SSO gateway. Source: https://docs.langflow.org/external-authentication (reviewed 2026-11-20).

- SCIM provisioning: Not researched.

- Compliance attestations: Not researched.

- Security controls: Documented. Current advisory review: Langflow 1.13.3 extends custom-component policy enforcement to background workflows and invalidates native sessions after password changes. Upgrade to 1.13.3 or later and retest component policy and session revocation. Scope: Upstream 1.13.3 release notes; no consolidated affected-version range is asserted.. Source: https://github.com/langflow-ai/langflow/releases/tag/v1.13.3 (reviewed 2026-11-18).

- Air-gapped operation: Not researched.

- Deployment artifact / IaC: Documented. The Docker guide provides an official image and Compose with persistent PostgreSQL, plus custom-image options. Preserve database/flow data and configure secrets, access controls and model-provider credentials. Scope: Upstream self-hosted deployment; not a provider support or production-readiness certification. Source: https://docs.langflow.org/deployment-docker (reviewed 2026-11-14).

- Minimum footprint / sizing: Documented. Langflow OSS lists a dual-core CPU and 2 GB RAM minimum, recommending a multi-core CPU and at least 4 GB. This covers the Python application; external databases, model inference, workflow concurrency and HA are additional. Scope: Langflow OSS application baseline; model and production topology excluded. Source: https://docs.langflow.org/get-started-installation (reviewed 2026-11-20).

- License pricing: Documented. The reviewed upstream core is MIT licensed. Optional services, bundled dependencies, commercial distributions and support can have separate terms. This records the source licence, not an infrastructure or operating-cost estimate. Model weights and external model APIs require their own review. Scope: Upstream core licence; commercial features, support and hosting priced separately. Source: https://raw.githubusercontent.com/langflow-ai/langflow/main/LICENSE (reviewed 2026-11-20).

- Operating cost estimate: Not researched.

- Release snapshot / cadence: Documented. Langflow v1.13.3 was published November 17, 2026. It applies custom-component policy to background workflows, invalidates native sessions after password changes, updates security dependencies, and aligns Docker and wheel packaging. Scope: Upstream Langflow self-hosted release. Source: https://github.com/langflow-ai/langflow/releases/tag/v1.13.3 (reviewed 2026-11-18).




### Security and lifecycle reviews

Selected editorial reviews only; no flag does not mean safe or supported.

- Langflow 1.13.3 policy and session hardening (security-advisory): Langflow 1.13.3 extends custom-component policy enforcement to background workflows and invalidates native sessions after password changes. Scope: Upstream 1.13.3 release notes; no consolidated affected-version range is asserted. Conditions: Applicability depends on custom components, workflow execution mode, and native authentication. Action: Upgrade to 1.13.3 or later and retest component policy and session revocation. Source: https://github.com/langflow-ai/langflow/releases/tag/v1.13.3; reviewed 2026-11-18.




## LibreChat Self-Hosted

Self-hosted multi-model AI chat, agent and retrieval platform with documented Railway and DigitalOcean deployments.

- Listing: https://launchinyourcloud.com/directory/librechat/
- Publisher listing: No
- Entry kind: product
- Documented cloud IDs: railway, digitalocean
- Operating model: Customer-operated
- Infrastructure IDs: servers
- License: MIT
- Licensing model: open-source
- Installation: Official Railway one-click template or Docker on a DigitalOcean Ubuntu Droplet
- Architecture: Containerized web and API services with application databases, retrieval services and customer-configured model endpoints
- Operations: Customer manages the Railway project or DigitalOcean Droplet, application secrets, model credentials, data services, backups, scaling and upgrades. Railway operates its platform; Droplet users also maintain the operating system and Docker runtime.
- Product website: https://www.librechat.ai/
- Source review: 2026-11-15; cloud review: 2026-11-15; licensing review: 2026-11-15; infrastructure review: 2026-11-15

### Scope and qualifications

The Railway path runs in the customer's project on platform-managed infrastructure; the DigitalOcean path uses a customer-operated Droplet. External model, embedding, search and tool providers receive data according to customer configuration. Review generated secrets, public networking, persistent data services and image updates before production use.

LibreChat is MIT-licensed. The licenses and commercial terms of configured model providers, tools and external services are separate.

LibreChat runs persistent application, database and retrieval containers in a Railway project or on a DigitalOcean Droplet.


### Category fit

- ai-applications-rag: Provides self-hosted multi-model chat, agents, file-backed retrieval and tool use over customer-selected model and data services.

- ai-tools: Provides a self-hosted interface and orchestration layer for model endpoints, agents, MCP tools, authentication and administrative controls.

### Deployment boundaries


- railway: Customer platform workspace; Customer-operated; control plane: not-assessed. Source: https://www.librechat.ai/docs/remote/railway (checked 2026-11-15). 

- digitalocean: Customer cloud account; Customer-operated; control plane: not-assessed. Source: https://www.librechat.ai/docs/remote/digitalocean (checked 2026-11-15). 


### Cloud deployment evidence

- LibreChat one-click deployment on Railway (railway): https://www.librechat.ai/docs/remote/railway
  LibreChat documents creating a Railway account, reviewing the template's three containers and environment variables, deploying the services and accessing the resulting LibreChat URL in the customer's Railway project.
  Source type: vendor-documentation; reviewed: 2026-11-15; vendor support: not-assessed.

- LibreChat deployment on DigitalOcean (digitalocean): https://www.librechat.ai/docs/remote/digitalocean
  The upstream guide creates an Ubuntu Droplet in the customer DigitalOcean project, configures a user and firewall, then links to the Docker installation steps. The example is for a small personal deployment, not a production sizing guarantee.
  Source type: vendor-documentation; reviewed: 2026-11-15; vendor support: not-assessed.


### Community and maintenance

LibreChat documents a community of hundreds of contributors and maintains its implementation in a public upstream repository. Contributor history establishes active project participation; it does not establish support for a specific Railway template release.

Reviewed 2026-11-15.

- LibreChat project and community: https://www.librechat.ai/about

- LibreChat contributor history: https://github.com/LibreChat-AI/LibreChat/graphs/contributors

### Sources

- LibreChat Railway deployment: https://www.librechat.ai/docs/remote/railway

- LibreChat project, self-hosting and MIT license: https://www.librechat.ai/about

- LibreChat upstream repository: https://github.com/LibreChat-AI/LibreChat

- LibreChat contributor history: https://github.com/LibreChat-AI/LibreChat/graphs/contributors

- LibreChat deployment on DigitalOcean: https://www.librechat.ai/docs/remote/digitalocean

### Licensing sources

- https://www.librechat.ai/about

### Infrastructure sources

- https://www.librechat.ai/docs/remote/railway

- https://www.librechat.ai/docs/remote/digitalocean


### Buyer diligence


Research coverage: 4/12 fields (33%); 4 documented, 0 unresolved. Coverage is not a feature score.

Unrecorded fields are not researched, not a statement that the feature is unavailable. Sources checked — unresolved means the cited material did not establish an answer. Hosted-service certifications do not automatically cover customer installations.


- Marketplace delivery: Not researched.

- Cloud commitment eligibility: Not researched.

- Single sign-on: Documented. The official self-hosted authentication guide documents OAuth2/OpenID Connect and SAML, with separately configured LDAP and local accounts. Configure the selected provider, registration policy, account linking and administrator recovery; external model credentials are a separate authentication surface. Scope: Self-hosted LibreChat user authentication. Source: https://www.librechat.ai/docs/configuration/authentication (reviewed 2026-11-20).

- SCIM provisioning: Not researched.

- Compliance attestations: Not researched.

- Security controls: Not researched.

- Air-gapped operation: Not researched.

- Deployment artifact / IaC: Documented. The Railway guide describes its repository deploy button, three-container configuration and required environment variables. Review persistence and optional integrations before production; the one-click recipe does not establish a minimum production footprint. Scope: LibreChat Railway deployment guide; moving template and configured features. Source: https://www.librechat.ai/docs/remote/railway (reviewed 2026-11-16).

- Minimum footprint / sizing: Documented. The Docker guide lists 1 vCPU and 1 GiB RAM minimum for LibreChat containers, recommending at least 2 GB allocated to Docker with the default backing services. MongoDB, search, RAG/pgvector, storage and HA need sizing; locally hosted models require additional compute and often GPUs. Scope: Official Docker starting profile; model inference and production HA excluded. Source: https://www.librechat.ai/docs/local/docker (reviewed 2026-11-20).

- License pricing: Documented. The reviewed upstream core is MIT licensed. Optional services, bundled dependencies, commercial distributions and support can have separate terms. This records the source licence, not an infrastructure or operating-cost estimate. Model weights and external model APIs require their own review. Scope: Upstream core licence; commercial features, support and hosting priced separately. Source: https://raw.githubusercontent.com/danny-avila/LibreChat/main/LICENSE (reviewed 2026-11-20).

- Operating cost estimate: Not researched.

- Release snapshot / cadence: Not researched.







## MaxKB Community Edition

Open-source knowledge-base, RAG and agent-workflow platform deployable into an Alibaba Cloud account.

- Listing: https://launchinyourcloud.com/directory/maxkb-community/
- Publisher listing: No
- Entry kind: product
- Documented cloud IDs: alibaba
- Operating model: Customer-operated
- Infrastructure IDs: servers
- License: GNU General Public License version 3 for Community Edition. Professional and Enterprise editions add separately licensed features and support.
- Licensing model: open-source
- Installation: Create the documented MaxKB Community Edition service instance through Alibaba Cloud Compute Nest, or use the publisher-documented Alibaba Cloud Marketplace image on a customer ECS server.
- Architecture: Customer-operated MaxKB application with PostgreSQL and pgvector storage on an Alibaba Cloud ECS host
- Operations: Customer manages the ECS host, VPC and security-group exposure, administrator credentials, persistent PostgreSQL data, backups, upgrades, capacity and credentials for model and tool integrations.
- Product website: https://www.maxkb.cn/
- Source review: 2026-11-19; cloud review: 2026-11-16; licensing review: 2026-11-16; infrastructure review: 2026-11-16

### Scope and qualifications

The Compute Nest deployment requires RAM permissions with full access to ECS, VPC and ROS plus Compute Nest user access; review and reduce deployment privileges after provisioning where practical. Community Edition is single-tenant, x86-64 only and supports single-node or cold-standby deployment rather than Enterprise clustering. RBAC, SSO, audit logs, open API access, conversation-user authorization and several enterprise integrations are reserved for paid editions. Current upstream v2 installation guidance calls for at least 4 CPU cores, 8 GB RAM and 100 GB of disk, exposes the web service on port 8080 and supplies default administrator credentials that must be changed immediately. The publisher recommends offline installation for production and requires a database backup before upgrades. Public or separately hosted model, embedding, reranking and tool providers can receive application data even when MaxKB itself runs in the customer account. The current upstream v2 container image and persistence paths differ from the legacy English guide; check the release packaged by the cloud installer before applying upstream configuration or upgrades. MaxKB v2.10.9-lts fixes multiple cross-workspace access and deletion flaws, stored XSS with account-takeover impact, custom-tool secret exposure, SSRF paths including possible cloud-credential theft, anonymous provider API-key disclosure and unauthorized conversation or knowledge-base access. Upgrade the LTS line and retest workspace, tool and provider boundaries.

This listing is scoped to the GPLv3 Community Edition. Professional and Enterprise editions add commercially licensed X-Pack capabilities, deployment options and vendor support; their entitlements are not implied by this classification.

Alibaba Compute Nest provisions the Community Edition on customer-selected ECS capacity, VPC, VSwitch, zone and region. The application and its persistent PostgreSQL/pgvector data run as provisioned server workloads rather than a serverless or vendor-managed service.


### Category fit

- ai-applications-rag: Self-hosted knowledge ingestion, document vectorization, RAG, question-answering and agent-workflow platform with configurable model providers.

- ai-tools: Customer-hosted agent and workflow platform with knowledge bases, MCP and tool integrations, and configurable language, embedding, reranking and multimodal models.

### Deployment boundaries


- alibaba: Customer cloud account; Customer-operated; control plane: not-assessed. Source: https://help.aliyun.com/en/compute-nest/use-cases/maxkb-community-edition-service-instance-deployment-document (checked 2026-11-16). Source: https://docs.maxkb.pro/installation/aliyun/ (checked 2026-11-16). 


### Cloud deployment evidence

- MaxKB Community Edition service instance in Alibaba Compute Nest (alibaba): https://help.aliyun.com/en/compute-nest/use-cases/maxkb-community-edition-service-instance-deployment-document
  Alibaba Cloud documents creation of a MaxKB Community Edition service instance in a customer-selected region, zone, VPC and VSwitch using customer-selected ECS capacity and billing. The deployed application is reached through the ECS public address on port 8080. The workflow requires broad ECS, VPC, ROS and Compute Nest permissions and returns default administrator credentials that must be changed.
  Source type: provider-guide; reviewed: 2026-11-16; vendor support: not-assessed.

- MaxKB image deployment from Alibaba Cloud Marketplace (alibaba): https://docs.maxkb.pro/installation/aliyun/
  The publisher documents selecting the MaxKB image while purchasing an Alibaba Cloud server, starting the image and opening port 8080 in the customer's security group. It recommends at least 4 CPU cores and 8 GB RAM and instructs administrators to replace the default password after first login.
  Source type: vendor-documentation; reviewed: 2026-11-16; vendor support: not-assessed.


### Community and maintenance

The official repository publishes an active versioned release stream and public contribution, code-of-conduct and security workflows. This establishes public project maintenance but does not certify any adopter's hosting topology.

Reviewed 2026-11-16.

- Upstream project and contribution workflow: https://github.com/1Panel-dev/MaxKB

- Current upstream releases: https://github.com/1Panel-dev/MaxKB/releases

### Sources

- Alibaba Compute Nest Community Edition deployment: https://help.aliyun.com/en/compute-nest/use-cases/maxkb-community-edition-service-instance-deployment-document

- Publisher Alibaba Cloud Marketplace deployment guide: https://docs.maxkb.pro/installation/aliyun/

- Upstream platform capabilities and technical stack: https://github.com/1Panel-dev/MaxKB

- Online container installation and requirements: https://docs.fit2cloud.com/maxkb/installation/online_installtion/

- Production-oriented offline installation and upgrade guidance: https://docs.fit2cloud.com/maxkb/installation/offline_installtion/

- Community, Professional and Enterprise edition comparison: https://www.maxkb.cn/price

- Community Edition GPLv3 license: https://github.com/1Panel-dev/MaxKB/blob/v2/LICENSE

- MaxKB v2.10.9-lts release: https://github.com/1Panel-dev/MaxKB/releases/tag/v2.10.9-lts

- MaxKB GHSA-5545-9jgw-jg2g: https://github.com/1Panel-dev/MaxKB/security/advisories/GHSA-5545-9jgw-jg2g

- MaxKB GHSA-mgph-cpg8-q442: https://github.com/1Panel-dev/MaxKB/security/advisories/GHSA-mgph-cpg8-q442

### Licensing sources

- https://github.com/1Panel-dev/MaxKB/blob/v2/LICENSE

- https://www.maxkb.cn/price

### Infrastructure sources

- https://help.aliyun.com/en/compute-nest/use-cases/maxkb-community-edition-service-instance-deployment-document

- https://docs.fit2cloud.com/maxkb/installation/online_installtion/

- https://github.com/1Panel-dev/MaxKB


### Buyer diligence


Research coverage: 8/12 fields (67%); 8 documented, 0 unresolved. Coverage is not a feature score.

Unrecorded fields are not researched, not a statement that the feature is unavailable. Sources checked — unresolved means the cited material did not establish an answer. Hosted-service certifications do not automatically cover customer installations.


- Marketplace delivery: Documented. Alibaba Cloud documents a Community Edition Compute Nest service that creates an ECS-based deployment in the customer’s selected network. The vendor’s separately priced Marketplace Professional Edition is a different entitlement; its pricing does not price this Community listing. Scope: Provider-documented Community Compute Nest service; separate from the Professional Marketplace offer. Source: https://help.aliyun.com/en/compute-nest/use-cases/maxkb-community-edition-service-instance-deployment-document (reviewed 2026-11-16). Supporting source: Edition pricing and Professional Marketplace distinction — https://www.maxkb.cn/price.

- Cloud commitment eligibility: Not researched.

- Single sign-on: Explicitly unsupported. The edition matrix excludes LDAP, OIDC, CAS, OAuth2 and SAML2 single sign-on from Community Edition and includes them in Professional/Enterprise. This does not establish SCIM provisioning. Scope: Community Edition only; commercial edition entitlements differ. Source: https://www.maxkb.cn/price (reviewed 2026-11-16).

- SCIM provisioning: Not researched.

- Compliance attestations: Not researched.

- Security controls: Documented. The edition matrix reserves full RBAC and system-operation logs for Professional/Enterprise. The online installer publishes default administrator credentials; replace them and configure access controls before exposing an instance. Current review: v2.10.9-lts fixes cross-workspace file and SKILL access/deletion, stored XSS with account-takeover impact, custom-tool secret exposure, multiple SSRF paths, anonymous provider API-key disclosure and unauthorized conversation or knowledge-base access. Upgrade and retest workspace and provider boundaries. Scope: MaxKB Community LTS; the release does not publish one universal affected-version range. Source: https://github.com/1Panel-dev/MaxKB/releases/tag/v2.10.9-lts (reviewed 2026-11-19). Supporting source: Prior security-controls source — https://www.maxkb.cn/price. Supporting source: Online installation and default account — https://docs.maxkb.pro/installation/online_installtion/. Supporting source: GHSA-5545-9jgw-jg2g — https://github.com/1Panel-dev/MaxKB/security/advisories/GHSA-5545-9jgw-jg2g. Supporting source: GHSA-mgph-cpg8-q442 — https://github.com/1Panel-dev/MaxKB/security/advisories/GHSA-mgph-cpg8-q442.

- Air-gapped operation: Documented. The v2 guide documents transferring a Community offline-installer archive, configuring it locally and applying offline upgrades after backups. That confirms offline application installation; models, embeddings, integrations and their dependencies must also be available inside the isolated environment. Scope: MaxKB v2 Community offline installation and upgrades; not a claim that external model APIs work disconnected. Source: https://docs.fit2cloud.com/maxkb/installation/offline_installtion/ (reviewed 2026-11-16).

- Deployment artifact / IaC: Documented. The current v2 guide uses registry.fit2cloud.com/maxkb/maxkb and persists data at /opt/maxkb. The English legacy guide uses different images/data paths; do not mix those configurations or assume compatibility with an older cloud package. Select and test a matching release. Scope: Current upstream v2 container example; untagged image is not an immutable release. Source: https://docs.fit2cloud.com/maxkb/installation/online_installtion/ (reviewed 2026-11-16). Supporting source: Versioned offline-installer example — https://docs.fit2cloud.com/maxkb/installation/offline_installtion/.

- Minimum footprint / sizing: Documented. The current v2 guide specifies at least 4 CPU cores, 8 GB RAM and 100 GB disk, with Docker 26+ recommended or the bundled Docker. Community package architecture and single-node/cold-standby limits follow the edition matrix; model hosting needs separate sizing. Scope: Community v2 application baseline, excluding model/GPU services and workload-specific capacity. Source: https://docs.fit2cloud.com/maxkb/installation/offline_installtion/ (reviewed 2026-11-16). Supporting source: Edition architecture/deployment matrix — https://www.maxkb.cn/price.

- License pricing: Documented. This listing is scoped to the GPLv3 Community Edition. Professional and Enterprise editions add commercially licensed X-Pack capabilities, deployment options and vendor support; their entitlements are not implied by this classification. Scope: MaxKB Community Edition: software license terms; infrastructure and operating costs are separate. Source: https://github.com/1Panel-dev/MaxKB/blob/v2/LICENSE (reviewed 2026-11-16). Supporting source: Edition and license conditions — https://www.maxkb.cn/price.

- Operating cost estimate: Not researched.

- Release snapshot / cadence: Documented. MaxKB v2.10.9-lts was published November 19, 2026. It fixes cross-workspace authorization, stored XSS, secret exposure, SSRF, provider-key disclosure and unauthorized conversation or knowledge-base access issues. Scope: Upstream MaxKB Community LTS release; commercial editions and deployment packages may differ. Source: https://github.com/1Panel-dev/MaxKB/releases/tag/v2.10.9-lts (reviewed 2026-11-19). Supporting source: Prior reviewed release context — https://github.com/1Panel-dev/MaxKB/releases/tag/v2.10.8-lts.




### Security and lifecycle reviews

Selected editorial reviews only; no flag does not mean safe or supported.

- MaxKB v2.10.9-lts security update (security-advisory): MaxKB v2.10.9-lts fixes cross-workspace authorization, stored XSS, secret exposure, SSRF, provider-key disclosure and unauthorized data access. Scope: MaxKB Community LTS; the release does not publish one universal affected-version range. Conditions: Workspace membership, custom tools, provider configuration, anonymous access and reachable metadata or internal services determine applicability. Action: Upgrade to v2.10.9-lts or later and retest workspace, tool, provider and anonymous-access boundaries. Source: https://github.com/1Panel-dev/MaxKB/releases/tag/v2.10.9-lts; reviewed 2026-11-19. Supporting source: GHSA-5545-9jgw-jg2g — https://github.com/1Panel-dev/MaxKB/security/advisories/GHSA-5545-9jgw-jg2g. Supporting source: GHSA-mgph-cpg8-q442 — https://github.com/1Panel-dev/MaxKB/security/advisories/GHSA-mgph-cpg8-q442.




## Onyx Self-Hosted

Self-hosted enterprise search and AI assistant platform with connectors and retrieval infrastructure.

- Listing: https://launchinyourcloud.com/directory/onyx/
- Publisher listing: No
- Entry kind: product
- Documented cloud IDs: aws, azure, gcp, digitalocean
- Operating model: Customer-operated
- Infrastructure IDs: servers
- License: MIT Community Edition; Business and Enterprise features use separate commercial terms.
- Licensing model: open-source
- Installation: Docker Compose on a VM or official Kubernetes and Terraform deployment
- Architecture: Customer-operated web/API services, background workers, PostgreSQL, Redis, search and object storage
- Operations: Customer manages application services, databases, search, object storage, connectors, model credentials, backups, scaling and upgrades.
- Product website: https://www.onyx.app/
- Source review: 2026-11-14; cloud review: 2026-11-14; licensing review: 2026-11-14; infrastructure review: 2026-11-14

### Scope and qualifications

This entry covers self-hosted Onyx, not a vendor-hosted environment. External LLM, embedding and connector services receive data according to the customer's configuration. Business and Enterprise features and support require separate commercial entitlements.

Onyx Community Edition is MIT-licensed. Business and Enterprise features, images and support have separate commercial terms.

Onyx runs web services, workers and persistent database, cache, search and object-storage dependencies on VMs or Kubernetes.


### Category fit

- ai-applications-rag: Provides enterprise search, connectors and retrieval-augmented AI assistants over customer-selected knowledge sources.

- ai-tools: Provides self-hosted indexing, retrieval, model integration and agent services for building internal AI experiences.

### Deployment boundaries


- aws: Customer cloud account; Customer-operated; control plane: not-assessed. Source: https://docs.onyx.app/deployment/cloud/aws/ec2 (checked 2026-11-14). 

- azure: Customer cloud account; Customer-operated; control plane: not-assessed. Source: https://docs.onyx.app/deployment/cloud/azure (checked 2026-11-14). 

- gcp: Customer cloud account; Customer-operated; control plane: not-assessed. Source: https://docs.onyx.app/deployment/cloud/gcp (checked 2026-11-14). 

- digitalocean: Customer cloud account; Customer-operated; control plane: not-assessed. Source: https://docs.onyx.app/deployment/cloud/digital_ocean (checked 2026-11-14). 


### Cloud deployment evidence

- Onyx on AWS (aws): https://docs.onyx.app/deployment/cloud/aws/ec2
  Onyx documents self-hosting on a customer EC2 instance and links a production EKS architecture using AWS managed dependencies.
  Source type: vendor-documentation; reviewed: 2026-11-14; vendor support: not-assessed.

- Onyx on Azure (azure): https://docs.onyx.app/deployment/cloud/azure
  Onyx documents self-hosting on an Azure VM and an AKS Terraform path using customer Azure managed services.
  Source type: vendor-documentation; reviewed: 2026-11-14; vendor support: not-assessed.

- Onyx on Google Cloud (gcp): https://docs.onyx.app/deployment/cloud/gcp
  Onyx documents Compute Engine and production GKE deployment paths in the customer's Google Cloud project.
  Source type: vendor-documentation; reviewed: 2026-11-14; vendor support: not-assessed.

- Onyx on DigitalOcean (digitalocean): https://docs.onyx.app/deployment/cloud/digital_ocean
  Onyx documents installing its self-hosted stack on a customer DigitalOcean Droplet.
  Source type: vendor-documentation; reviewed: 2026-11-14; vendor support: not-assessed.


### Community and maintenance

Onyx has sustained, substantive code participation from multiple contributors. Release 5.1.2 on November 12, 2026 records continuing maintenance, including permission-handling and retrieval fixes. Contributor and release histories establish maintained project participation; they do not establish that paid Business or Enterprise features are included in the MIT Community Edition.

Reviewed 2026-11-14.

- Reviewed maintenance release v5.1.2: https://github.com/onyx-dot-app/onyx/releases/tag/v5.1.2

- Upstream contributor history: https://github.com/onyx-dot-app/onyx/graphs/contributors

### Sources

- AWS EC2 deployment: https://docs.onyx.app/deployment/cloud/aws/ec2

- Azure deployment: https://docs.onyx.app/deployment/cloud/azure

- Google Cloud deployment: https://docs.onyx.app/deployment/cloud/gcp

- DigitalOcean deployment: https://docs.onyx.app/deployment/cloud/digital_ocean

- Community license: https://github.com/onyx-dot-app/onyx/blob/main/LICENSE

- Reviewed maintenance release v5.1.2: https://github.com/onyx-dot-app/onyx/releases/tag/v5.1.2

- Upstream contributor history: https://github.com/onyx-dot-app/onyx/graphs/contributors

- Current Business and Enterprise edition boundaries: https://docs.onyx.app/deployment/enterprise_edition.md

### Licensing sources

- https://github.com/onyx-dot-app/onyx/blob/main/LICENSE

- https://docs.onyx.app/deployment/enterprise_edition.md

### Infrastructure sources

- https://docs.onyx.app/deployment/cloud/aws/ec2

- https://docs.onyx.app/deployment/cloud/gcp


### Buyer diligence


Research coverage: 6/12 fields (50%); 6 documented, 0 unresolved. Coverage is not a feature score.

Unrecorded fields are not researched, not a statement that the feature is unavailable. Sources checked — unresolved means the cited material did not establish an answer. Hosted-service certifications do not automatically cover customer installations.


- Marketplace delivery: Not researched.

- Cloud commitment eligibility: Not researched.

- Single sign-on: Documented. The upstream Enterprise section advertises Google OAuth, OIDC and SAML sign-in. The README does not establish every Community-versus-Enterprise protocol entitlement; confirm the selected self-hosted edition and identity configuration. Scope: Enterprise feature statement; exact protocol entitlement not fully established. Source: https://github.com/onyx-dot-app/onyx (reviewed 2026-11-14).

- SCIM provisioning: Edition-dependent. The upstream Enterprise section describes group synchronization and SCIM user provisioning. Confirm the licensed self-hosted implementation and supported lifecycle operations rather than assuming Community includes them. Scope: Onyx Enterprise feature statement. Source: https://github.com/onyx-dot-app/onyx (reviewed 2026-11-14).

- Compliance attestations: Not researched.

- Security controls: Not researched.

- Air-gapped operation: Not researched.

- Deployment artifact / IaC: Documented. The official Docker guide uses the upstream deployment/docker_compose directory and an environment template. The default image tag moves; replace example storage credentials and configure the selected inference/storage services. Scope: Upstream self-hosted deployment; not a provider support or production-readiness certification. Source: https://docs.onyx.app/deployment/local/docker (reviewed 2026-11-14). Deployment artifact: https://github.com/onyx-dot-app/onyx/tree/main/deployment/docker_compose. Artifact reference: main (moving branch).

- Minimum footprint / sizing: Documented. The vendor gives Lite minimum 2 vCPUs/2 GB RAM/10 GB disk and Standard minimum 4 vCPUs/10 GB RAM/32 GB plus roughly 2.5 times indexed data. Its Standard per-container example totals about 12 CPUs/24 GB. These describe different profiles; connector/index volume, model services and HA require additional sizing. Scope: Vendor Lite/Standard profiles; per-container and single-instance recommendations differ. Source: https://docs.onyx.app/deployment/getting_started/resourcing.md (reviewed 2026-11-20).

- License pricing: Documented. Core code is MIT licensed; ee directories are under the Onyx Enterprise License. Enterprise features need their own entitlement and quote. The MIT core licence does not establish all-edition feature access, support or model/hosting prices. Scope: Upstream core licence; commercial features, support and hosting priced separately. Source: https://raw.githubusercontent.com/onyx-dot-app/onyx/main/LICENSE (reviewed 2026-11-20).

- Operating cost estimate: Not researched.

- Release snapshot / cadence: Documented. The publisher-designated GitHub release checked was v5.1.2, published 2026-11-12. This is a release snapshot, not a promised cadence, support designation or approval to substitute it into a provider package. Scope: Upstream onyx-dot-app/onyx release record; commercial editions and deployment packages may differ. Source: https://github.com/onyx-dot-app/onyx/releases/tag/v5.1.2 (reviewed 2026-11-14).







## Open WebUI Self-Hosted

Self-hosted AI chat and knowledge interface for local model servers and cloud APIs.

- Listing: https://launchinyourcloud.com/directory/open-webui/
- Publisher listing: No
- Entry kind: product
- Documented cloud IDs: aws, azure, gcp
- Operating model: Customer-operated
- Infrastructure IDs: servers
- License: Open WebUI license based on BSD-style terms with additional branding restrictions and defined exceptions; enterprise terms may be required for rebranding.
- Licensing model: source-available
- Installation: Official Docker image, Python installation or Kubernetes deployment.
- Architecture: Customer-operated web application and conversation storage connected to model providers
- Operations: Customer runs application storage, authentication, provider configuration, updates and backups.
- Product website: https://openwebui.com/
- Source review: 2026-11-18; cloud review: 2026-11-12; licensing review: 2026-11-10; infrastructure review: 2026-11-10

### Scope and qualifications

The UI sends messages to the model endpoint selected for a conversation. Local Ollama or vLLM endpoints can keep inference on customer infrastructure; cloud APIs, embeddings, tools and plugins have separate data flows. The current license restricts branding removal in larger deployments, with small-deployment, permission and enterprise-license exceptions; review the actual terms. Open WebUI v0.13.4 replaces Direct Connections and Integrations with Tether and migrates image settings. Deployments using Terminals server connections require Terminals 0.4.0 or later; upgrading that server from 0.3.x requires manual migration. Open Terminal and Tether terminal connections are unaffected by this migration. Tether introduces user-device and local-command paths that require explicit trust review.

The current Open WebUI license adds branding restrictions to BSD-style terms, with small-deployment and permission/license exceptions. Classified here as source-available rather than treating the current version as unmodified BSD-3-Clause.

The application runs as a Docker/Python server or Kubernetes workload with persistent conversation/user storage. Local inference is a separate server or accelerator workload, depending on the chosen model.


### Category fit

- ai-applications-rag: Deployable team chat and model-access interface with retrieval features. It is an interface and application layer, not an included license or runtime for every connected model.

- ai-tools: Self-hosted AI chat and knowledge interface for local model servers and cloud APIs.

### Deployment boundaries


- aws: Customer cloud account; Customer-operated; control plane: not-assessed. Source: https://docs.openwebui.com/deployment/aws-ecs/ (checked 2026-11-12). 

- azure: Customer cloud account; Customer-operated; control plane: not-assessed. Source: https://docs.openwebui.com/deployment/azure-container-apps/ (checked 2026-11-12). 

- gcp: Customer cloud account; Customer-operated; control plane: not-assessed. Source: https://docs.openwebui.com/deployment/google-cloud-run/ (checked 2026-11-12). 


### Cloud deployment evidence

- Open WebUI deployment on AWS ECS / Fargate (aws): https://docs.openwebui.com/deployment/aws-ecs/
  Project deployment instructions cover the application on AWS ECS / Fargate with external database, Redis and file storage. Model inference is separate; review model endpoints, egress and WebSocket/runtime limits.
  Source type: vendor-documentation; reviewed: 2026-11-12; vendor support: not-assessed.

- Open WebUI deployment on Azure Container Apps (azure): https://docs.openwebui.com/deployment/azure-container-apps/
  Project deployment instructions cover the application on Azure Container Apps with external database, Redis and file storage. Model inference is separate; review model endpoints, egress and WebSocket/runtime limits.
  Source type: vendor-documentation; reviewed: 2026-11-12; vendor support: not-assessed.

- Open WebUI deployment on Google Cloud Run (gcp): https://docs.openwebui.com/deployment/google-cloud-run/
  Project deployment instructions cover the application on Google Cloud Run with external database, Redis and file storage. Model inference is separate; review model endpoints, egress and WebSocket/runtime limits.
  Source type: vendor-documentation; reviewed: 2026-11-12; vendor support: not-assessed.


### Sources

- Installation options: https://docs.openwebui.com/getting-started/quick-start/

- Local and cloud model connections: https://docs.openwebui.com/getting-started/quick-start/connect-a-provider/

- Current license and branding conditions: https://docs.openwebui.com/license/

- Licensing and edition terms: https://github.com/open-webui/open-webui/blob/main/LICENSE

- Open WebUI v0.13.4 release: https://github.com/open-webui/open-webui/releases/tag/v0.13.4

### Licensing sources

- https://github.com/open-webui/open-webui/blob/main/LICENSE

- https://docs.openwebui.com/license/

### Infrastructure sources

- https://docs.openwebui.com/getting-started/quick-start/


### Buyer diligence


Research coverage: 5/12 fields (42%); 5 documented, 0 unresolved. Coverage is not a feature score.

Unrecorded fields are not researched, not a statement that the feature is unavailable. Sources checked — unresolved means the cited material did not establish an answer. Hosted-service certifications do not automatically cover customer installations.


- Marketplace delivery: Not researched.

- Cloud commitment eligibility: Not researched.

- Single sign-on: Documented. Upstream documents OAuth/OIDC and trusted-header authentication. Only one generic OIDC provider is configured through OPENID_PROVIDER_URL at a time; trusted-header use requires an isolated trusted proxy. Scope: Upstream authentication configuration; review provider and session settings. Source: https://docs.openwebui.com/features/authentication-access/auth/sso/ (reviewed 2026-11-14).

- SCIM provisioning: Not researched.

- Compliance attestations: Not researched.

- Security controls: Not researched.

- Air-gapped operation: Not researched.

- Deployment artifact / IaC: Documented. The upstream Compose example includes Open WebUI and Ollama with persistent volumes. Its defaults use moving image references and an empty application secret; configure production credentials and pin versions. Scope: Upstream example; model resources and service exposure require separate review. Source: https://github.com/open-webui/open-webui/blob/main/docker-compose.yaml (reviewed 2026-11-14). Deployment artifact: https://github.com/open-webui/open-webui/blob/main/docker-compose.yaml. Artifact reference: main (moving branch).

- Minimum footprint / sizing: Documented. The performance guide shows a container example reserving 2 CPUs/4 GB RAM with limits of 4 CPUs/8 GB. Multi-user deployments need PostgreSQL; scaled replicas also require shared services. Local inference, embeddings, extraction, vector storage and HA are additional. This is an example allocation, not a validated 100-user or 1-TB minimum. Scope: Application-container example; full stack and workload sizing excluded. Source: https://docs.openwebui.com/troubleshooting/performance/ (reviewed 2026-11-20). Supporting source: Shared infrastructure for scaled deployments — https://docs.openwebui.com/enterprise/deployment/.

- License pricing: Documented. Current Open WebUI has custom terms restricting branding changes, with exceptions including deployments of at most 50 end users in a rolling 30-day period, written permission or an enterprise licence. A 100-user deployment cannot assume the small-deployment branding exception. Model and infrastructure costs are separate. Scope: Upstream core licence; commercial features, support and hosting priced separately. Source: https://raw.githubusercontent.com/open-webui/open-webui/main/LICENSE (reviewed 2026-11-20).

- Operating cost estimate: Not researched.

- Release snapshot / cadence: Documented. Open WebUI v0.13.4 was published November 17, 2026. Deployments using Terminals server connections require Terminals 0.4.0 or later; upgrading that server from 0.3.x requires manual migration. Open Terminal and Tether terminal connections are unaffected by this migration. The release also replaces Direct Connections and Integrations with Tether and migrates image settings. Scope: Upstream Open WebUI self-hosted release; the Terminals version and migration requirements apply only to deployments using Terminals server connections. Source: https://github.com/open-webui/open-webui/releases/tag/v0.13.4 (reviewed 2026-11-18).







## RAGFlow

Document processing and retrieval platform for building RAG and agentic knowledge applications.

- Listing: https://launchinyourcloud.com/directory/ragflow/
- Publisher listing: No
- Entry kind: product
- Documented cloud IDs: aws, azure, gcp, alibaba, railway
- Operating model: Customer-operated
- Infrastructure IDs: servers
- License: Apache License 2.0 for upstream RAGFlow; models, data assets and dependencies retain their own terms.
- Licensing model: open-source
- Installation: Versioned upstream Docker Compose, provider-specific VM packages or community deployment templates; select a path compatible with the intended RAGFlow release.
- Architecture: Customer-operated document/retrieval application with deployment-specific stateful storage and configured model services
- Operations: Customer manages document storage, indexing, app services, access controls, model endpoints, backups and upgrades.
- Product website: https://ragflow.io/
- Source review: 2026-11-12; cloud review: 2026-11-12; licensing review: 2026-11-10; infrastructure review: 2026-11-10

### Scope and qualifications

Deployment options include customer AWS, Azure and Google Cloud infrastructure, Alibaba Compute Nest, and a Railway project. The evidence trail identifies the publishers and template versions. Hosting RAGFlow does not automatically self-host its LLM, embeddings or reranker; choose private or local model endpoints where required.

Apache License 2.0 for upstream RAGFlow; models, data assets and dependencies retain their own terms. This classification covers the reviewed upstream edition; external services and dependencies retain their own terms.

Customer-operated document/retrieval application with deployment-specific stateful storage and configured model services. This is a customer-operated server/container deployment; portability does not establish named cloud-provider support.


### Category fit

- ai-applications-rag: Document understanding, knowledge ingestion and retrieval-augmented/agentic question answering with configurable LLM, embedding and reranking endpoints.

- ai-tools: Self-hosted document ingestion, retrieval and context infrastructure for AI applications.

### Deployment boundaries


- aws: Customer cloud account; Customer-operated; control plane: not-assessed. Source: https://github.com/stdapi-ai/samples/tree/main/getting_started_ragflow (checked 2026-11-12). 

- azure: Customer cloud account; Customer-operated; control plane: not-assessed. Source: https://www.cloudimg.co.uk/guides/ragflow-on-ubuntu-24-04-azure/ (checked 2026-11-12). 

- gcp: Customer cloud account; Customer-operated; control plane: not-assessed. Source: https://www.techlatest.net/support/ragflow_support/gcp_gettingstartedguide/index.html (checked 2026-11-12). 

- alibaba: Customer cloud account; Customer-operated; control plane: not-assessed. Source: https://help.aliyun.com/en/compute-nest/use-cases/ragflow-community-edition-service-instance-deployment-document (checked 2026-11-12). 

- railway: Customer platform workspace; Customer-operated; control plane: not-assessed. Source: https://railway.com/deploy/ragflow-ai (checked 2026-11-12). Source: https://github.com/gridalpha/ragflow-railway/blob/main/Dockerfile (checked 2026-11-12). 


### Cloud deployment evidence

- Community RAGFlow deployment on AWS ECS Fargate through stdapi.ai (aws): https://github.com/stdapi-ai/samples/tree/main/getting_started_ragflow
  The sample provisions the RAGFlow application and workers, OpenSearch, Aurora PostgreSQL, Valkey, S3 and a Bedrock gateway in the customer AWS account. Source inspection confirms the v0.26.4 Python image pin; it does not establish compatibility with the 1.0 Go rewrite. This community integration also requires the separately billed stdapi.ai gateway and Bedrock access. Review its documented networking, client-security, availability and cost limitations before use.
  Source type: community-template; reviewed: 2026-11-12; vendor support: not-assessed. Edition: Community / upstream open-source; third-party distribution where noted. Versions: 0.26.4.

- cloudimg RAGFlow Azure VM deployment and operations guide (azure): https://www.cloudimg.co.uk/guides/ragflow-on-ubuntu-24-04-azure/
  The image publisher documents an Azure VM in the customer subscription running the complete Compose stack, including Elasticsearch, MySQL, MinIO, Redis and a local embedding service. Its guide pins RAGFlow v0.26.4, not 1.0. This is a third-party repackaged image with additional support charges, not an upstream-endorsed distribution. Confirm image access, current patch/migration requirements, TLS configuration and the separately configured chat-model endpoint.
  Source type: third-party-package; reviewed: 2026-11-12; vendor support: not-assessed. Edition: Community / upstream open-source; third-party distribution where noted. Versions: 0.26.4.

- Techlatest RAGFlow VM deployment in a customer Google Cloud project (gcp): https://www.techlatest.net/support/ragflow_support/gcp_gettingstartedguide/index.html
  The image publisher documents selecting the customer project, deployment service account, zone, network and Compute Engine VM, then completing RAGFlow initialization and configuring its bundled Ollama models. This is a third-party packaged deployment; the guide does not establish a specific current RAGFlow release or 1.0 compatibility. Confirm the image version, support charges, model licenses, resource sizing and network exposure before use.
  Source type: third-party-package; reviewed: 2026-11-12; vendor support: not-assessed. Edition: Community / upstream open-source; third-party distribution where noted.

- RAGFlow Community deployment on Alibaba Cloud Compute Nest (alibaba): https://help.aliyun.com/en/compute-nest/use-cases/ragflow-community-edition-service-instance-deployment-document
  Alibaba Cloud documents provisioning a RAGFlow Community service instance with customer ECS, VPC and vSwitch parameters. This is distinct from the managed AI-Native Database Service integration. Review image version, access controls, model endpoints and the broad example RAM permissions before production.
  Source type: provider-guide; reviewed: 2026-11-12; vendor support: not-assessed. Edition: Community / upstream open-source; third-party distribution where noted.

- Community RAGFlow Railway application stack (railway): https://railway.com/deploy/ragflow-ai
  This community template deploys the application, Elasticsearch, MySQL, Redis, a local embedding service and object storage into the customer Railway project. Its linked Dockerfile pins upstream RAGFlow v0.27.1 with a Railway startup/configuration wrapper. The template is not proof of 1.0 compatibility or upstream support; review migrations and the separately configured chat-model endpoint. Railway manages the underlying infrastructure.
  Source type: community-template; reviewed: 2026-11-12; vendor support: not-assessed. Edition: Community / upstream open-source; third-party distribution where noted. Versions: 0.27.1.

- Railway deployment image source and release pin (railway): https://github.com/gridalpha/ragflow-railway/blob/main/Dockerfile
  Source inspection confirms the v0.27.1 base image and the custom configuration, nginx and startup wrapper used by the linked Railway template. The upstream 1.0 release changes the architecture and requires its documented migration path; do not substitute a new image tag without reviewing the complete stack.
  Source type: community-template; reviewed: 2026-11-12; vendor support: not-assessed. Edition: Community / upstream open-source; third-party distribution where noted. Versions: 0.27.1.


### Community and maintenance

Current release notes document continued 2026 releases and dozens of named contributors across successive versions; the upstream README records a 1.0.0 release on 6 November 2026.

Reviewed 2026-11-10.

- Current releases and contributors: https://github.com/infiniflow/ragflow/releases

- Release chronology and deployment instructions: https://raw.githubusercontent.com/infiniflow/ragflow/main/README.md

### Sources

- Upstream project and capabilities: https://github.com/infiniflow/ragflow

- Current upstream license: https://raw.githubusercontent.com/infiniflow/ragflow/main/LICENSE

- RAGFlow 1.0 release and migration requirements: https://ragflow.io/docs/v1.0.0/release_notes

### Licensing sources

- https://raw.githubusercontent.com/infiniflow/ragflow/main/LICENSE

### Infrastructure sources

- https://github.com/infiniflow/ragflow


### Buyer diligence


Research coverage: 5/12 fields (42%); 4 documented, 1 unresolved. Coverage is not a feature score.

Unrecorded fields are not researched, not a statement that the feature is unavailable. Sources checked — unresolved means the cited material did not establish an answer. Hosted-service certifications do not automatically cover customer installations.


- Marketplace delivery: Not researched.

- Cloud commitment eligibility: Not researched.

- Single sign-on: Sources checked — unresolved. A vendor maintainer identifies OAuth2/OIDC SSO in Enterprise, while the reviewed v1.0.0 Go deployment reference does not establish a supported federation configuration for that OSS release. Older OAuth configuration examples must not be assumed compatible with the new backend. Confirm edition, release and identity-provider configuration. Scope: RAGFlow v1.0 Go OSS versus Enterprise federation; current entitlement/configuration unresolved. Source: https://raw.githubusercontent.com/infiniflow/ragflow/v1.0.0/docker/README.md (reviewed 2026-11-20). Supporting source: Vendor-maintainer Enterprise SSO statement — https://github.com/infiniflow/ragflow/issues/15882.

- SCIM provisioning: Not researched.

- Compliance attestations: Not researched.

- Security controls: Not researched.

- Air-gapped operation: Not researched.

- Deployment artifact / IaC: Documented. The v1.0.0 Compose file includes a separate dependency configuration and reads its image selection from the release environment. Use the matching release configuration; this artifact does not establish a migration path or compatibility for older provider packages. Scope: Upstream v1.0.0 Compose; not evidence of compatibility with the listed 0.x provider packages. Source: https://github.com/infiniflow/ragflow/blob/v1.0.0/docker/docker-compose.yml (reviewed 2026-11-14). Deployment artifact: https://github.com/infiniflow/ragflow/blob/v1.0.0/docker/docker-compose.yml. Artifact reference: v1.0.0. Supporting source: Versioned RAGFlow 1.0.0 quickstart — https://ragflow.io/docs/v1.0.0/.

- Minimum footprint / sizing: Documented. The 1.0 quickstart recommends a starting point of 4 CPU cores, 16 GB RAM and 50 GB free disk. Models, document engines and concurrency can require more; this does not size the older third-party cloud packages. Scope: Upstream 1.0 quickstart; provider packages may target 0.x. Source: https://ragflow.io/docs/v1.0.0/ (reviewed 2026-11-14).

- License pricing: Documented. The reviewed upstream core is Apache-2.0 licensed. Optional services, bundled dependencies, commercial distributions and support can have separate terms. This records the source licence, not an infrastructure or operating-cost estimate. Model weights and external model APIs require their own review. Scope: Upstream core licence; commercial features, support and hosting priced separately. Source: https://raw.githubusercontent.com/infiniflow/ragflow/main/LICENSE (reviewed 2026-11-20).

- Operating cost estimate: Not researched.

- Release snapshot / cadence: Documented. The publisher-designated GitHub release checked was v1.0.0, published 2026-11-06. This is a release snapshot, not a promised cadence, support designation or approval to substitute it into a provider package. Scope: Upstream infiniflow/ragflow release record; commercial editions and deployment packages may differ. Source: https://github.com/infiniflow/ragflow/releases/tag/v1.0.0 (reviewed 2026-11-14).






