
# AI observability & LLMOps on AWS

Compare documented AI observability & LLMOps deployments on AWS, including operating responsibility, licensing and deployment evidence.

Canonical page: https://launchinyourcloud.com/clouds/aws/ai-observability/
Catalog revision: 170de95bb7d832897d269051ad4500713acf96a62f3d01bc1d4e649679a56fde




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.


## Provider guidance

- Account boundary: AWS account
- Provider group: cloud-infrastructure
- Provider classification: Navigation group for the documented infrastructure offering; read each application path for its actual account boundary. Source: https://docs.aws.amazon.com/accounts/latest/reference/getting-started.html (checked 2026-11-20).
- Network: VPC
- Identity and access: IAM roles and policies
- Installation guidance: CloudFormation or Terraform; review stack changes and requested IAM permissions.
- Guidance source: https://docs.aws.amazon.com/AWSCloudFormation/latest/UserGuide/Welcome.html
- Guidance reviewed: 2026-11-10


### Private offers

AWS Marketplace private offers can contain negotiated pricing and terms for a buyer account. Confirm the product delivery model and your agreement; commitment credit is not established by a directory badge.
Source: https://docs.aws.amazon.com/marketplace/latest/buyerguide/buyer-private-offers-subscribing.html (reviewed 2026-11-14).

### Private connectivity

PrivateLink provides private endpoints to supported services and resources across VPC/account boundaries. Confirm that the selected product exposes a compatible endpoint service.
Source: https://docs.aws.amazon.com/vpc/latest/privatelink/what-is-privatelink.html (reviewed 2026-11-14).

### Cross-account operations

Use an IAM role with scoped permissions and a third-party external ID where applicable. Review the trust principal, support access and revocation separately from installation access.
Source: https://docs.aws.amazon.com/IAM/latest/UserGuide/id_roles_common-scenarios_third-party.html (reviewed 2026-11-14).

### Government environments

AWS GovCloud (US) is a distinct isolated-region environment. A commercial-region product deployment does not establish availability or certification there.
Source: https://docs.aws.amazon.com/govcloud-us/latest/UserGuide/whatis.html (reviewed 2026-11-14).

The listings below have evidence for Amazon Web Services. Read each cloud evidence note for the applicable package, edition and support qualifications.

3 listings.


## Arize Phoenix Self-Hosted

AI observability and evaluation server for traces, datasets and application experiments.

- Listing: https://launchinyourcloud.com/directory/phoenix/
- Publisher listing: No
- Entry kind: product
- Documented cloud IDs: aws, azure, gcp, railway, render
- Operating model: Customer-operated
- Infrastructure IDs: servers
- License: Elastic License 2.0; includes restrictions on providing the software as a hosted/managed service and on license-key circumvention. This is source-available.
- Licensing model: source-available
- Installation: Use the official self-hosted Docker/Compose or Kubernetes deployment instructions and configure persistent storage and instrumentation endpoints.
- Architecture: Customer-operated trace/evaluation application with persistent trace and dataset storage
- Operations: Customer manages trace capture, storage, authentication, retention, evaluation providers, backups and upgrades.
- Product website: https://arize.com/phoenix/
- Source review: 2026-11-12; cloud review: 2026-11-12; licensing review: 2026-11-10; infrastructure review: 2026-11-10

### Scope and qualifications

The self-hosted product is distinct from commercial Arize offerings. Traces may contain prompts and outputs; evaluation jobs can call separately configured external model services even when trace storage is local.

Elastic License 2.0; includes restrictions on providing the software as a hosted/managed service and on license-key circumvention. This is source-available. This classification covers the reviewed upstream edition; external services and dependencies retain their own terms.

Customer-operated trace/evaluation application with persistent trace and dataset storage. This is a customer-operated server/container deployment; portability does not establish named cloud-provider support.


### Category fit

- ai-observability: AI trace collection, evaluation, datasets, experiments and prompt analysis for instrumented model applications.

- ai-tools: AI development and operations tooling for tracing and evaluation rather than a model-serving runtime.

### Deployment boundaries


- aws: Customer cloud account; Customer-operated; control plane: not-assessed. Source: https://arize.com/docs/phoenix/self-hosting (checked 2026-11-12). 

- azure: Customer cloud account; Customer-operated; control plane: not-assessed. Source: https://arize.com/docs/phoenix/self-hosting (checked 2026-11-12). 

- gcp: Customer cloud account; Customer-operated; control plane: not-assessed. Source: https://arize.com/docs/phoenix/self-hosting (checked 2026-11-12). 

- railway: Customer platform workspace; Customer-operated; control plane: not-assessed. Source: https://arize.com/docs/phoenix/self-hosting (checked 2026-11-12). 

- render: Customer platform workspace; Customer-operated; control plane: not-assessed. Source: https://arize.com/docs/phoenix/self-hosting (checked 2026-11-12). 


### Cloud deployment evidence

- Phoenix self-hosting deployment options (aws, azure, gcp, railway): https://arize.com/docs/phoenix/self-hosting
  The current Phoenix self-hosting index includes a Railway one-click template as well as AWS CloudFormation, Azure ARM and Google Cloud Run. These deploy the Phoenix application, not the separately licensed Arize AX product. A Railway project is customer-controlled while Railway manages its infrastructure.
  Source type: vendor-documentation; reviewed: 2026-11-12; vendor support: not-assessed.

- Phoenix self-hosted Render Blueprint (render): https://arize.com/docs/phoenix/self-hosting
  The vendor self-hosting index offers a Render one-click Blueprint for Phoenix alongside its Railway, AWS, Azure and Google Cloud options. This is the self-hosted Phoenix application, separate from Arize AX; customer workspace control does not imply ownership of the platform infrastructure.
  Source type: vendor-documentation; reviewed: 2026-11-12; vendor support: not-assessed.


### Community and maintenance

Arize's June 2026 maintainer account describes actual participation outside the core team: it attributes OIDC authentication to users needing Keycloak/Cognito and credits contributors with expanding framework integrations and OpenInference. The upstream release history shows continuing maintenance. This concerns concrete community involvement rather than repository star counts; it does not certify any named customer's deployment model.

Reviewed 2026-11-10.

- Maintainer account of user-contributed authentication and integrations: https://arize.com/blog/phoenix-10k/

- Current Phoenix maintenance releases: https://github.com/Arize-ai/phoenix/releases

### Sources

- Official installation and runtime: https://arize.com/docs/phoenix/self-hosting

- Upstream project and capabilities: https://github.com/Arize-ai/phoenix

- Current upstream license: https://raw.githubusercontent.com/Arize-ai/phoenix/main/LICENSE

### Licensing sources

- https://raw.githubusercontent.com/Arize-ai/phoenix/main/LICENSE

### Infrastructure sources

- https://arize.com/docs/phoenix/self-hosting


### 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: Documented. Phoenix documents OAuth2 authorization-code login with OIDC provider discovery, configured alongside application authentication and secret settings. Configure the trusted provider, callback and account policy; client API keys remain a separate surface. Scope: Self-hosted Phoenix UI OAuth2/OIDC. Source: https://arize.com/docs/phoenix/deployment/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 upstream Phoenix repository provides self-hosting/container documentation for the observability server. Match persistence and authentication settings to the chosen release; instrumented applications and model services remain separate. Scope: Upstream self-hosted deployment; not a provider support or production-readiness certification. Source: https://github.com/arize-ai/phoenix (reviewed 2026-11-14). Deployment artifact: https://github.com/arize-ai/phoenix. Artifact reference: repository default branch (moving; select matching application release).

- Minimum footprint / sizing: Sources checked — unresolved. The self-hosting guide documents deployment, storage/database configuration and trace retention, but does not establish a complete CPU/RAM minimum for a specified workload. Size span ingestion, retention, database, evaluation/model dependencies and redundant service instances. Hosted Arize AX sizing is a separate offering. Scope: Phoenix self-hosted application and data services; numeric workload sizing unresolved. Source: https://arize.com/docs/phoenix/self-hosting/deploying-phoenix.md (reviewed 2026-11-20).

- License pricing: Documented. Phoenix uses Elastic License 2.0, including limits on offering substantial hosted/managed functionality and bypassing licence controls. This is not an unrestricted Apache/MIT licence or a hosted-service price; review commercial service/support terms separately. Scope: Upstream core licence; commercial features, support and hosting priced separately. Source: https://raw.githubusercontent.com/Arize-ai/phoenix/main/LICENSE (reviewed 2026-11-20).

- Operating cost estimate: Not researched.

- Release snapshot / cadence: Documented. The publisher-designated GitHub release checked was arize-phoenix-v21.1.0, published 2026-11-13. This is a release snapshot, not a promised cadence, support designation or approval to substitute it into a provider package. Scope: Upstream Arize-ai/phoenix release record; commercial editions and deployment packages may differ. Source: https://github.com/Arize-ai/phoenix/releases/tag/arize-phoenix-v21.1.0 (reviewed 2026-11-14).







## Langfuse Self-Hosted

Self-hosted LLM observability, evaluations and prompt-management platform.

- Listing: https://launchinyourcloud.com/directory/langfuse/
- Publisher listing: No
- Entry kind: product
- Documented cloud IDs: aws, azure, gcp, railway, render
- Operating model: Customer-operated
- Infrastructure IDs: servers
- License: MIT core; designated enterprise directories and add-on capabilities use separate license terms or keys.
- Licensing model: open-source
- Installation: Docker Compose for a single VM; Helm/Kubernetes and official AWS, Azure and Google Cloud Terraform modules.
- Architecture: Application and worker containers with PostgreSQL, ClickHouse, Redis/Valkey and object storage
- Operations: Customer operates the deployment, persistence, access policies, retention, backups and scaling.
- Product website: https://langfuse.com/
- Source review: 2026-11-18; cloud review: 2026-11-12; licensing review: 2026-11-10; infrastructure review: 2026-11-10

### Scope and qualifications

Traces can contain prompts, outputs and other sensitive application data, so configure capture and retention deliberately. Self-hosted trace storage does not move the underlying application model into the same account. Optional model-based evaluators or the gateway use configured providers, which may be external. The single-VM Compose example does not provide production high availability or backups. Langfuse v4.82.0 fixes an SSRF path in GitHub Enterprise SSO and moves audit logs to ClickHouse. Review outbound SSO connectivity and audit-log storage during upgrade.

The main codebase uses MIT terms outside designated enterprise directories. Enterprise code requires a separate valid license, so the directory treats the combined product as open-core.

The documented deployment uses application/worker containers and persistent database, cache and object-storage services. Host/cluster capacity and database operations remain part of the self-hosted system.


Cloud evidence scope: Documented provider-specific deployment

### Category fit

- ai-observability: LLM/agent tracing, evaluation and prompt management deployed with customer-controlled storage. Evaluation and optional gateway providers must be scoped independently.

- ai-tools: Self-hosted LLM observability, evaluations and prompt-management platform.

- observability: Application/model traces and experiment visibility for AI workloads. Infrastructure metrics and broader application monitoring require separately selected tooling.

### Deployment boundaries


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

- azure: Customer cloud account; Customer-operated; control plane: not-assessed. Source: https://langfuse.com/self-hosting/deployment/azure (checked 2026-11-12). 

- gcp: Customer cloud account; Customer-operated; control plane: not-assessed. Source: https://langfuse.com/self-hosting/deployment/gcp (checked 2026-11-12). 

- railway: Customer platform workspace; Customer-operated; control plane: not-assessed. Source: https://langfuse.com/self-hosting/deployment/railway (checked 2026-11-12). 

- render: Customer platform workspace; Customer-operated; control plane: not-assessed. Source: https://langfuse.com/self-hosting/deployment/render (checked 2026-11-12). 


### Cloud deployment evidence

- Langfuse customer cloud deployment on AWS (aws): https://langfuse.com/self-hosting/deployment/aws
  The vendor documents deployment of the self-hosted Langfuse application and its backing services in the customer account or project, including provider-specific infrastructure. This is distinct from Langfuse Cloud; database, object storage, model calls and enterprise licensing must be configured for the chosen topology.
  Source type: vendor-documentation; reviewed: 2026-11-12; vendor support: not-assessed.

- Langfuse customer cloud deployment on Azure (azure): https://langfuse.com/self-hosting/deployment/azure
  The vendor documents deployment of the self-hosted Langfuse application and its backing services in the customer account or project, including provider-specific infrastructure. This is distinct from Langfuse Cloud; database, object storage, model calls and enterprise licensing must be configured for the chosen topology.
  Source type: vendor-documentation; reviewed: 2026-11-12; vendor support: not-assessed.

- Langfuse customer cloud deployment on Google Cloud (gcp): https://langfuse.com/self-hosting/deployment/gcp
  The vendor documents deployment of the self-hosted Langfuse application and its backing services in the customer account or project, including provider-specific infrastructure. This is distinct from Langfuse Cloud; database, object storage, model calls and enterprise licensing must be configured for the chosen topology.
  Source type: vendor-documentation; reviewed: 2026-11-12; vendor support: not-assessed.

- Langfuse v3 community Railway deployment (railway): https://langfuse.com/self-hosting/deployment/railway
  The vendor documents a community-maintained Railway template for the complete Langfuse v3 stack. This establishes that version-specific application deployment; the same page explicitly distinguishes the v4 alternatives. The vendor does not systematically test the community template. Review the template services, image tags, database timezones and upgrade path.
  Source type: vendor-documentation; reviewed: 2026-11-12; vendor support: not-assessed.

- Langfuse v4 community Render Blueprint (render): https://langfuse.com/self-hosting/deployment/render
  The vendor documents a community-maintained v4 Blueprint with separate web and worker services, PostgreSQL, Key Value, ClickHouse and MinIO. It creates the full application in the customer Render workspace. Review service plans, persistent disks, object-storage exposure and recovery procedures; community support is best-effort.
  Source type: vendor-documentation; reviewed: 2026-11-12; vendor support: not-assessed.


### Sources

- Docker Compose on a VM: https://langfuse.com/self-hosting/deployment/docker-compose

- Self-hosting architecture and provider dependencies: https://langfuse.com/self-hosting

- AWS account Terraform deployment: https://langfuse.com/self-hosting/deployment/aws

- Azure account Terraform deployment: https://langfuse.com/self-hosting/deployment/azure

- Google Cloud account Terraform deployment: https://langfuse.com/self-hosting/deployment/gcp

- Core and enterprise licensing: https://github.com/langfuse/langfuse/blob/main/LICENSE

- Licensing and edition terms: https://github.com/langfuse/langfuse/blob/main/ee/LICENSE

- Langfuse v3 community Railway deployment: https://langfuse.com/self-hosting/deployment/railway

- Langfuse v4 community Render Blueprint: https://langfuse.com/self-hosting/deployment/render

- Langfuse v4.82.0 release: https://github.com/langfuse/langfuse/releases/tag/v4.82.0

### Licensing sources

- https://github.com/langfuse/langfuse/blob/main/LICENSE

- https://github.com/langfuse/langfuse/blob/main/ee/LICENSE

### Infrastructure sources

- https://langfuse.com/self-hosting/deployment/docker-compose

- https://langfuse.com/self-hosting/deployment/aws

- https://langfuse.com/self-hosting/deployment/azure

- https://langfuse.com/self-hosting/deployment/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 self-hosted guide documents OAuth/OIDC providers including Keycloak, Okta, Entra ID and a custom issuer. Configure account linking, email verification and access policy; SCIM and separately licensed enterprise features require their own review. Scope: Self-hosted OAuth/OIDC sign-in. Source: https://langfuse.com/self-hosting/security/authentication-and-sso (reviewed 2026-11-20).

- SCIM provisioning: Not researched.

- Compliance attestations: Not researched.

- Security controls: Documented. Current advisory review: Langfuse v4.82.0 fixes a server-side request forgery path in GitHub Enterprise SSO. Upgrade to v4.82.0 or later and restrict outbound SSO connectivity. Scope: Upstream v4.82.0 release; the vendor did not publish an affected-version range in the reviewed note.. Source: https://github.com/langfuse/langfuse/releases/tag/v4.82.0 (reviewed 2026-11-18).

- Air-gapped operation: Not researched.

- Deployment artifact / IaC: Documented. The upstream Compose recipe defines application services and backing data services. Pin a compatible Langfuse release and review credentials, persistent storage and exposed ports. Scope: Upstream Compose recipe; supporting-service operations remain customer responsibilities. Source: https://github.com/langfuse/langfuse/blob/main/docker-compose.yml (reviewed 2026-11-14). Deployment artifact: https://github.com/langfuse/langfuse/blob/main/docker-compose.yml. Artifact reference: main (moving branch).

- Minimum footprint / sizing: Documented. The vendor recommends at least 2 CPUs and 4 GB RAM for application containers and at least two Web instances for HA. Worker, ClickHouse, PostgreSQL, Redis and blob storage capacity are additional. Confirm per-container resources and retention/ingestion volume rather than treating this as the whole-stack footprint. Scope: Langfuse Web/Worker resource recommendation; backing services additional. Source: https://langfuse.com/self-hosting/deployment/infrastructure/containers (reviewed 2026-11-20).

- License pricing: Documented. Core code is MIT licensed, while ee directories, including application and worker enterprise paths, use a separate enterprise licence. Optional commercial features and support need separate terms; underlying databases, object storage and model services are additional. Scope: Upstream core licence; commercial features, support and hosting priced separately. Source: https://raw.githubusercontent.com/langfuse/langfuse/main/LICENSE (reviewed 2026-11-20).

- Operating cost estimate: Not researched.

- Release snapshot / cadence: Documented. Langfuse v4.82.0 was published November 17, 2026. It fixes an SSRF path in GitHub Enterprise SSO and moves audit logs to ClickHouse. Scope: Upstream Langfuse self-hosted release. Source: https://github.com/langfuse/langfuse/releases/tag/v4.82.0 (reviewed 2026-11-18).




### Security and lifecycle reviews

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

- Langfuse GitHub Enterprise SSO SSRF fix (security-advisory): Langfuse v4.82.0 fixes a server-side request forgery path in GitHub Enterprise SSO. Scope: Upstream v4.82.0 release; the vendor did not publish an affected-version range in the reviewed note. Conditions: Requires the affected GitHub Enterprise SSO configuration and a reachable outbound target. Action: Upgrade to v4.82.0 or later and restrict outbound SSO connectivity. Source: https://github.com/langfuse/langfuse/releases/tag/v4.82.0; reviewed 2026-11-18.




## MLflow Tracking Server Self-Hosted

Self-hosted experiment tracking and LLM/agent observability with customer-controlled metadata and artifacts.

- Listing: https://launchinyourcloud.com/directory/mlflow/
- Publisher listing: No
- Entry kind: product
- Documented cloud IDs: aws, azure, gcp
- Operating model: Customer-operated
- Infrastructure IDs: servers
- License: Apache 2.0 for upstream MLflow; connected model providers and managed distributions have separate terms.
- Licensing model: open-source
- Installation: Python MLflow server command, official Docker Compose bundle or Kubernetes Helm deployment.
- Architecture: Customer-operated tracking/API server with configurable metadata database and artifact storage
- Operations: Customer operates server access, metadata and artifact stores, retention, backups, scaling and version upgrades.
- Product website: https://mlflow.org/
- Source review: 2026-11-12; cloud review: 2026-11-12; licensing review: 2026-11-10; infrastructure review: 2026-11-10

### Scope and qualifications

Traces can capture model inputs, outputs and intermediate metadata. Point clients at the customer-hosted server and review recording/redaction settings. Applications, evaluators and optional model integrations may still call external providers; hosting the tracking server does not host all training or inference runtimes. Production database, storage and security configuration differ from a local quickstart.

The upstream self-hosted tracking server is Apache-2.0 licensed. This classification does not extend to separate managed MLflow distributions, connected model providers or their commercial services.

The tracking server runs as a Python service or container with metadata/artifact stores. A local SQLite example is not a production architecture; separately plan persistent database, storage and application-runtime infrastructure.


### Category fit

- ai-observability: Experiment tracking and LLM/agent tracing server. This entry describes customer-hosted tracking/observability services, not every model runtime, managed MLflow product or inference backend.

- ai-tools: Self-hosted experiment tracking and LLM/agent observability with customer-controlled metadata and artifacts.

- observability: Application/model traces and experiment visibility for AI workloads. Infrastructure metrics and broader application monitoring require separately selected tooling.

### Deployment boundaries


- aws: Customer cloud account; Customer-operated; control plane: not-assessed. Source: https://mlflow.org/docs/latest/self-hosting/deploy-to-cloud/aws/ (checked 2026-11-12). 

- azure: Customer cloud account; Customer-operated; control plane: not-assessed. Source: https://mlflow.org/docs/latest/self-hosting/deploy-to-cloud/azure/ (checked 2026-11-12). 

- gcp: Customer cloud account; Customer-operated; control plane: not-assessed. Source: https://mlflow.org/docs/latest/self-hosting/deploy-to-cloud/gcp/ (checked 2026-11-12). 


### Cloud deployment evidence

- Self-hosted MLflow deployment on AWS (aws): https://mlflow.org/docs/latest/self-hosting/deploy-to-cloud/aws/
  The project documents deployment of the MLflow server, metadata database and artifact storage on AWS. This does not refer to a managed MLflow edition or to model serving alone.
  Source type: vendor-documentation; reviewed: 2026-11-12; vendor support: not-assessed.

- Self-hosted MLflow deployment on Azure (azure): https://mlflow.org/docs/latest/self-hosting/deploy-to-cloud/azure/
  The project documents deployment of the MLflow server, metadata database and artifact storage on Azure. This does not refer to a managed MLflow edition or to model serving alone.
  Source type: vendor-documentation; reviewed: 2026-11-12; vendor support: not-assessed.

- Self-hosted MLflow deployment on Google Cloud (gcp): https://mlflow.org/docs/latest/self-hosting/deploy-to-cloud/gcp/
  The project documents deployment of the MLflow server, metadata database and artifact storage on Google Cloud. This does not refer to a managed MLflow edition or to model serving alone.
  Source type: vendor-documentation; reviewed: 2026-11-12; vendor support: not-assessed.


### Sources

- Self-hosting server and deployment methods: https://mlflow.org/docs/latest/self-hosting/

- LLM/agent tracing and captured data: https://mlflow.org/docs/latest/genai/tracing/

- Apache 2.0 license: https://github.com/mlflow/mlflow/blob/master/LICENSE.txt

### Licensing sources

- https://github.com/mlflow/mlflow/blob/master/LICENSE.txt

### Infrastructure sources

- https://mlflow.org/docs/latest/self-hosting/


### 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: Documented. The official guide documents the community-maintained mlflow-oidc-auth plugin and an SSO reverse-proxy pattern. This provides UI/API OIDC integration, but plugin maintenance, MLflow-version compatibility and client token behavior require separate review; it is not a claim of vendor-operated identity support. Scope: Officially documented community-plugin/proxy integration. Source: https://mlflow.org/docs/latest/self-hosting/security/sso/ (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 self-hosting guide links an upstream Compose bundle with MLflow, PostgreSQL and object storage, plus an official Helm option. Configure production credentials and persistence rather than retaining development defaults. Scope: Upstream self-hosted deployment; not a provider support or production-readiness certification. Source: https://mlflow.org/docs/latest/self-hosting/ (reviewed 2026-11-14). Deployment artifact: https://github.com/mlflow/mlflow/tree/master/docker-compose. Artifact reference: master Compose bundle (moving).

- Minimum footprint / sizing: Sources checked — unresolved. The self-hosted guide documents the UI/API tracking server, metadata backend and artifact-store configuration, but not a universal production CPU/RAM minimum. Size runs/traces, request concurrency, database, artifact storage and redundancy; starting the default local SQLite server does not establish production or HA capacity. Scope: Self-hosted tracking-server topology; numeric production sizing unresolved. Source: https://mlflow.org/docs/latest/self-hosting/architecture/tracking-server/ (reviewed 2026-11-20).

- 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. Scope: Upstream core licence; commercial features, support and hosting priced separately. Source: https://raw.githubusercontent.com/mlflow/mlflow/master/LICENSE.txt (reviewed 2026-11-20).

- Operating cost estimate: Not researched.

- Release snapshot / cadence: Documented. The publisher-designated GitHub release checked was v3.18.0, published 2026-11-04. This is a release snapshot, not a promised cadence, support designation or approval to substitute it into a provider package. Scope: Upstream mlflow/mlflow release record; commercial editions and deployment packages may differ. Source: https://github.com/mlflow/mlflow/releases/tag/v3.18.0 (reviewed 2026-11-14).






