AI tools & infrastructure on Azure
AI applications, model operations, inference and supporting vector databases.
20 documented options on Microsoft Azure. Provider coverage reflects reviewed evidence, not an exhaustive compatibility claim.
Read this page as JSON or Markdown.
| Listing | Documented clouds | Deployment and infrastructure | Licensing | Installation | Operations and limitations |
|---|---|---|---|---|---|
| Arize Phoenix Self-Hosted | AWS, Azure, Google Cloud, Railway, Render | Customer-operated Servers / containers | Source Available Elastic License 2.0; includes restrictions on providing the software as a hosted/managed service and on license-key circumvention. This is source-available. | Use the official self-hosted Docker/Compose or Kubernetes deployment instructions and configure persistent storage and instrumentation endpoints. | Customer manages trace capture, storage, authentication, retention, evaluation providers, backups and upgrades.Scope and evidenceThe 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.
Cloud evidence reviewed 2026-11-12 |
| Cloudera on Cloud | AWS, Azure, Google Cloud | Shared operations · vendor control plane Servers / containers | Commercial Only Commercial Cloudera subscription with cloud infrastructure charges paid separately. | Register the customer cloud environment and credentials, then provision Cloudera workload services through the control plane | Cloudera operates the service control plane; the customer owns cloud resources, credentials, data governance, network integration, workload sizing and shared operational responsibilities.Scope and evidenceThis entry covers Cloudera's public-cloud BYOC service, not a fully customer-operated software distribution. Workloads and data run in the customer's AWS, Azure or Google Cloud environment while Cloudera operates the control plane. Service availability and implementation details differ by provider, especially on Google Cloud, so confirm each required service before procurement. Cloudera November bulletins cover critical Brotli CVE-2026-111497 and compromised eMudhra certificates. Use the TSB matrices for the exact Runtime or Private Cloud Base component and fixed build rather than applying a generic platform version.
Cloud evidence reviewed 2026-11-17 |
| Dataiku DSS Cloud Stacks | AWS, Azure, Google Cloud | Customer-operated Servers / containers | Commercial Only Commercial Dataiku license or license token required for the full Cloud Stacks capabilities. | AWS CloudFormation, Azure ARM template or Google Cloud Infrastructure Manager with Terraform | Customer manages cloud accounts, Fleet Manager, IAM identities, networking, VM and Kubernetes capacity, storage, TLS, backups, upgrades and license configuration.Scope and evidenceThis entry covers Dataiku Cloud Stacks, not vendor-hosted Dataiku Cloud. Dataiku states that Cloud Stacks resources run in the customer's organization or tenant and that Dataiku does not have access to customer data. Fleet Manager requires broad cloud permissions and outbound access for images, license validation and configured identity services. Dataiku DSS Free Edition is separate from the fully featured licensed Cloud Stacks deployment.
Cloud evidence reviewed 2026-11-16 |
| Elastic Cloud on Kubernetes | AWS, Azure, Google Cloud | Customer-operated Servers / containers | Source Available Elastic License 2.0 distribution with free Basic capabilities; Enterprise licensing unlocks additional features and support. | Official ECK Helm chart or Kubernetes manifests on EKS, AKS or GKE | Customer manages Kubernetes, storage, resource sizing, security, snapshots, upgrades and availability; Elastic licensing controls advanced features and support.Scope and evidenceThis entry covers self-managed Elastic Cloud on Kubernetes, not Elastic Cloud Hosted or Elastic Cloud Enterprise. ECK automates application lifecycle but does not transfer cluster, storage, backup or upgrade responsibility to Elastic. The free Basic license and paid Enterprise feature set differ; optional Cloud Connect services create separate external dependencies. ECK 3.5.0 fixes service-token exposure, cross-namespace certificate injection, and retained cross-namespace credentials. CVE-2026-72648 has no workaround; after upgrading, rotate the affected service token as Elastic directs.
Cloud evidence reviewed 2026-11-17 |
| Hyperscience Hypercell | AWS, Azure, Google Cloud | Customer-operated Servers / containers | Commercial Only Commercial proprietary licence; product, user, database and support entitlements must be agreed for the selected installation. | Helm / Kubernetes operator; Kubernetes manifests | Customer operates the application and its cloud resources, including secrets, backups, monitoring and upgrades. Review any support access and configured external services; preserve application data, configuration and keys for migration or exit.Scope and evidenceThe Helm guide supplies complete AWS and GCP values, including application storage permissions. Registry integration alone is not a hosting badge. Hypercell application versions, Helm charts and Kubernetes support dates are separate. The October 2026 decoder advisory and the current v44 flow/block upgrade instructions must be checked before exposing document processing; hosted FedRAMP claims do not transfer to this installation. The vendor Azure reference diagram depicts the complete platform, Blob storage and PostgreSQL; its namespace label still says GKE, so validate the Azure-specific values with the vendor. Apply the fix for the installed Hypercell line and follow the documented sequential upgrades or temporary workarounds.
Cloud evidence reviewed 2026-11-19 |
| KNIME Business Hub | AWS, Azure, Google Cloud | Customer-operated Servers / containers | Commercial Only Commercial KNIME Business Hub license or Marketplace software subscription required. | Official AWS, Azure or Google Cloud Marketplace virtual-machine image | Customer manages the cloud account, virtual machine, networking, DNS, TLS, identity, storage, backups, capacity, upgrades and Business Hub license.Scope and evidenceThis entry covers customer-hosted KNIME Business Hub, not the open-source KNIME Analytics Platform or KNIME's managed Hub offering. Marketplace deployments use single VM images; production capacity, availability and recovery requirements must be assessed separately. Customer-provisioned Kubernetes installation requires the applicable Self Hosted Premium package. Cloud infrastructure charges are separate from KNIME software pricing.
Cloud evidence reviewed 2026-11-16 |
| Langflow Self-Hosted | AWS, Azure, Google Cloud | Customer-operated Servers / containers | Open Source / Open Core MIT for the upstream project; connected models, providers and bundles have their own terms. | Official Docker image or Docker Compose; Python installation and Kubernetes options are also documented. | Customer manages the application, database, access controls, installed components, backups and version upgrades.Scope and evidenceThe 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.
Cloud evidence reviewed 2026-11-12 |
| Langfuse Self-Hosted | AWS, Azure, Google Cloud, Railway, Render | Customer-operated Servers / containers | Open Source / Open Core MIT core; designated enterprise directories and add-on capabilities use separate license terms or keys. | Docker Compose for a single VM; Helm/Kubernetes and official AWS, Azure and Google Cloud Terraform modules. | Customer operates the deployment, persistence, access policies, retention, backups and scaling.Scope and evidenceTraces 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.
Cloud evidence reviewed 2026-11-12 |
| LiteLLM Proxy Self-Hosted | AWS, Azure, Google Cloud, Railway, Render | Customer-operated Servers / containers | Open Source / Open Core MIT outside the enterprise directory; enterprise code and features have separate license terms. | Official container image, Docker Compose or Kubernetes deployment; production infrastructure examples are documented. | Customer manages gateway availability, credentials, database, access policies, logs and upgrades, including pinning versions and verifying release artifacts with the documented version-appropriate Cosign method; model providers remain separate services.Scope and evidenceThe gateway forwards each request to its configured model provider. Self-hosting the proxy does not ensure prompts stay inside the account: external model APIs, guardrail services and logging integrations can receive data. Use appropriate self-hosted inference endpoints and explicitly review outbound integrations when required. For release integrity, release images and the Helm chart starting with v1.111.0-dev.3 use keyless Cosign signatures bound to LiteLLM's release workflow; earlier releases require the documented historical public keys. The October 21 incident affected PyPI v1.107.0 and Docker Hub/GHCR v1.107.0 plus mutable latest, main-latest and main-stable tags during the documented exposure windows. Check the incident's artifact digests and remediation instructions; the compromised images carried valid historical signatures, so signature verification alone does not establish safety. LiteLLM v1.110.0 changes authentication, guardrail, accounting, and S3 behavior. The 1.108.11 LTS patch also fixes Bedrock guarded_text handling and restricts os.environ team settings to proxy administrators; review both the selected release line and existing incident-response or image-signing guidance. LiteLLM v1.108.12 LTS fixes tool_policy allowing blocked /v1/messages tool names longer than 64 characters. The standard v1.110.1 UI image changes to slim nginx and removes curl while retaining wget and nc, so retest health probes and scripts that assume curl.
Cloud evidence reviewed 2026-11-12 |
| Milvus | AWS, Azure, Google Cloud | Customer-operated Servers / containers | Open Source / Open Core Apache License 2.0 for upstream Milvus. | Run the official standalone Docker or versioned Docker Compose deployment; distributed deployments have separate operational requirements. | Customer manages persistent volumes, collections, indexes, access controls, capacity, backups and upgrades.Scope and evidenceCurrent installation scripts can track a different image than a selected stable release; use versioned configurations for reproducibility. Model inference and embedding generation are separate workloads. Managed Zilliz offerings are outside this listing.
Cloud evidence reviewed 2026-11-12 |
| MLflow Tracking Server Self-Hosted | AWS, Azure, Google Cloud | Customer-operated Servers / containers | Open Source / Open Core Apache 2.0 for upstream MLflow; connected model providers and managed distributions have separate terms. | Python MLflow server command, official Docker Compose bundle or Kubernetes Helm deployment. | Customer operates server access, metadata and artifact stores, retention, backups, scaling and version upgrades.Scope and evidenceTraces 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.
Cloud evidence reviewed 2026-11-12 |
| n8n Self-Hosted | AWS, Azure, Google Cloud, DigitalOcean, Heroku, Hetzner | Customer-operated Servers / containers | Source Available Sustainable Use License for covered code; separate commercial Enterprise license for enterprise components | Docker Compose with persistent data; database and worker topology selected for workload | Customer manages servers/containers, secrets, database, backup/restore, scaling and upgrades.Scope and evidenceThis entry excludes n8n Cloud. The Sustainable Use License has use restrictions and should not be described as unrestricted open-source licensing. Enterprise components require the appropriate commercial license. External integrations, AI providers and optional assistant services require separate data-flow review. n8n@2.48.6 and the maintained n8n@1.123.99 line move official images to patched Node runtimes carrying a V8 process-abort fix. The 1.x image also moves to Node 24.22.3, while 2.x pins a patched Node 26.13.0. Update the selected line and verify custom images use the fixed runtime.
Cloud evidence reviewed 2026-11-12 |
| NVIDIA AI Enterprise | AWS, Azure, Google Cloud, Oracle Cloud, Alibaba Cloud, Tencent Cloud, Volcano Engine | Customer-operated Servers / containers | Commercial Only Commercial NVIDIA AI Enterprise subscription, consumption or perpetual license; most customer-cloud paths require BYOL. | NVIDIA AI Enterprise or GPU-optimized cloud VM images, standard instances with BYOL, or managed Kubernetes on supported cloud providers | Customer manages cloud GPU instances or Kubernetes, drivers, GPU Operator and container runtime, networking, storage, identities, NGC access, model services, monitoring, upgrades and license activation.Scope and evidenceThis entry covers customer-operated NVIDIA AI Enterprise software, not a managed AI service. A license is required and is not included with most standard, GPU-optimized VMI or managed-Kubernetes deployments; on-demand NVIDIA AI Enterprise VMIs and private offers use separate included-license processes. Not every component is available on every cloud or deployment model. GPU deployments are generally licensed per GPU, while CPU-only instances require an instance license. Tencent Cloud and ByteDance Volcano Engine have documented standard-instance BYOL paths. These guides require certified software and supported instance types; they do not establish that every GPU, component or provider image is eligible. Their per-instance wording differs from the general per-GPU licensing guidance, so confirm the applicable licensing unit. The November lifecycle matrix lists active 9.0/R620, 8.3/R595, and 7.9 LTSB/R580 branches with different end dates. NeMo Microservices support ended October 1, 2026, after 25.12 became the final GA release; select and patch each component branch independently.
Cloud evidence reviewed 2026-11-16 |
| Onyx Self-Hosted | AWS, Azure, Google Cloud, DigitalOcean | Customer-operated Servers / containers | Open Source / Open Core MIT Community Edition; Business and Enterprise features use separate commercial terms. | Docker Compose on a VM or official Kubernetes and Terraform deployment | Customer manages application services, databases, search, object storage, connectors, model credentials, backups, scaling and upgrades.Scope and evidenceThis 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.
Cloud evidence reviewed 2026-11-14 |
| Open WebUI Self-Hosted | AWS, Azure, Google Cloud | Customer-operated Servers / containers | Source Available Open WebUI license based on BSD-style terms with additional branding restrictions and defined exceptions; enterprise terms may be required for rebranding. | Official Docker image, Python installation or Kubernetes deployment. | Customer runs application storage, authentication, provider configuration, updates and backups.Scope and evidenceThe 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.
Cloud evidence reviewed 2026-11-12 |
| RAGFlow | AWS, Azure, Google Cloud, Alibaba Cloud, Railway | Customer-operated Servers / containers | Open Source / Open Core Apache License 2.0 for upstream RAGFlow; models, data assets and dependencies retain their own terms. | Versioned upstream Docker Compose, provider-specific VM packages or community deployment templates; select a path compatible with the intended RAGFlow release. | Customer manages document storage, indexing, app services, access controls, model endpoints, backups and upgrades.Scope and evidenceDeployment 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.
Cloud evidence reviewed 2026-11-12 |
| Red Hat OpenShift AI Self-Managed | AWS, Azure, Google Cloud, Oracle Cloud, IBM Cloud | Customer-operated Servers / containers | Commercial Only Commercial Red Hat OpenShift AI Self-Managed subscription; an eligible OpenShift subscription is also required. | OpenShift AI Operator through OperatorHub or the oc command-line interface | Customer manages the OpenShift cluster, storage, accelerators, identity, networking, model data, backups and upgrades within Red Hat's supported configuration matrix.Scope and evidenceThis entry covers OpenShift AI Self-Managed, not Red Hat's managed cloud-service add-on. Support depends on a compatible OpenShift Container Platform release, storage and accelerator matrix, and active subscriptions. Underlying OpenShift and cloud infrastructure are operated by the customer. RHSA-2026:95872 is a critical OpenShift AI 3.5.3 security update covering nine CVEs. Use Red Hat's package and deployment scope to determine affected components, then apply 3.5.3 or a later supported update.
Cloud evidence reviewed 2026-11-17 |
| SAS Viya Platform | AWS, Azure, Google Cloud | Customer-operated Servers / containers | Commercial Only Commercial SAS Viya licensing; deployment assets and a JWT license are delivered for the customer's software order. | Provision EKS, AKS or GKE with SAS Terraform projects, then deploy Kubernetes manifests generated through Kustomize and SAS automation | Customer manages Kubernetes, networking, ingress, storage, PostgreSQL, identity, sizing, backups, platform updates and licensed deployment assets.Scope and evidenceThe public SAS infrastructure repositories are Apache-licensed deployment tooling; SAS Viya itself is commercial software and requires licensed order assets and a JWT license. Kubernetes and PostgreSQL are mandatory. SAS states that its infrastructure-as-code projects satisfy most, not all, deployment requirements. The viya4-deployment project can patch the same manifest but does not move an installation to a different cadence, version or software offering. SAS November bulletins cover compromised eMudhra certificates and critical Brotli CVE-2026-111497 across specified Viya branches. Use the vendor matrices to identify the exact affected component and fixed maintenance level; do not treat a generic Viya version as sufficient.
Cloud evidence reviewed 2026-11-17 |
| Strategy Enterprise in Your Cloud | AWS, Azure, Google Cloud, STACKIT | Vendor-operated BYOC Servers / containers | Commercial Only Commercial Strategy Enterprise subscription; customer-cloud deployment is not available with the Standard plan. | Vendor-assisted, fully managed deployment into the customer's licensed AWS, Azure Google Cloud or STACKIT environment | Strategy administers and maintains the platform; the customer supplies the cloud environment, data sources, identity, network approvals and shared governance responsibilities.Scope and evidenceThis entry covers the Strategy Enterprise customer-cloud option, not Strategy Standard in the vendor's managed environment. Strategy describes the service as fully managed despite its deployment in a customer-licensed AWS, Azure Google Cloud or STACKIT environment, so it is classified as BYOC rather than fully customer-operated. Confirm contract-specific account ownership, cloud-resource billing, service availability, support boundaries and cloud-commitment eligibility before procurement.
Cloud evidence reviewed 2026-11-17 |
| vLLM Self-Hosted | AWS, Azure, Google Cloud | Customer-operated Servers / containers | Open Source / Open Core Apache 2.0 for vLLM; model weights and supporting dependencies have independent licenses and access conditions. | Official vLLM Docker image or supported Python installation; choose a compatible hardware/backend configuration. | Customer manages model acquisition, hardware capacity, endpoint protection, availability, upgrades and model/runtime compatibility.Scope and evidenceThis entry covers the inference server, not a hosted model service. It can run supported model weights on customer infrastructure, but those weights may need external downloads or access approval and are not licensed by vLLM. Review default usage telemetry and optional outbound integrations; hosting inference does not automatically create an air-gapped system. vLLM v0.34.0 removes Model Runner V1; configurations setting VLLM_USE_V2_MODEL_RUNNER=0 now fail at startup. It changes the default block size from 16 to 64, enables extensible KV cache by default with gpu_memory_utilization defaulting to 1.0, drops ROCm 7.2 wheels and images in favor of ROCm 10.0, requires huggingface_hub 2.0.0 or later, and removes QuTLASS/fp_quant and Marlin W4A8 support. Review the upstream breaking-change table and test model, quantization, memory, and accelerator compatibility before rollout.
Cloud evidence reviewed 2026-11-12 |
Deployment evidence and category fit
Arize Phoenix Self-Hosted
AI development and operations tooling for tracing and evaluation rather than a model-serving runtime.
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.
- Phoenix self-hosting deployment options ↗
Vendor / upstream documentation · Reviewed 2026-11-12. Vendor support commitment: not assessed; confirm the selected release and contract.
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.
Cloud evidence reviewed 2026-11-12. Full listing and qualifications
Cloudera on Cloud
Machine-learning and AI workload services are available within the Cloudera platform, subject to provider and subscription availability.
This entry covers Cloudera's public-cloud BYOC service, not a fully customer-operated software distribution. Workloads and data run in the customer's AWS, Azure or Google Cloud environment while Cloudera operates the control plane. Service availability and implementation details differ by provider, especially on Google Cloud, so confirm each required service before procurement. Cloudera November bulletins cover critical Brotli CVE-2026-111497 and compromised eMudhra certificates. Use the TSB matrices for the exact Runtime or Private Cloud Base component and fixed build rather than applying a generic platform version.
- Cloudera workloads in customer AWS, Azure and Google Cloud environments ↗
Vendor / upstream documentation · Reviewed 2026-11-17. Vendor support commitment: not assessed; confirm the selected release and contract.
Cloudera documents registering customer AWS, Azure and Google Cloud environments and credentials so its control plane can provision workload resources. Provider service availability differs and must be checked for the selected workload.
Cloud evidence reviewed 2026-11-17. Full listing and qualifications
Dataiku DSS Cloud Stacks
Provides customer-operated data science, machine-learning, generative-AI, model-deployment and AI-governance capabilities.
This entry covers Dataiku Cloud Stacks, not vendor-hosted Dataiku Cloud. Dataiku states that Cloud Stacks resources run in the customer's organization or tenant and that Dataiku does not have access to customer data. Fleet Manager requires broad cloud permissions and outbound access for images, license validation and configured identity services. Dataiku DSS Free Edition is separate from the fully featured licensed Cloud Stacks deployment.
- Dataiku DSS Cloud Stacks on Azure ↗
Vendor / upstream documentation · Reviewed 2026-11-16. Vendor support commitment: not assessed; confirm the selected release and contract.
Dataiku documents deploying Fleet Manager and its supporting resources through an Azure ARM template, then provisioning DSS instances and optional AKS elastic compute in the customer's subscription.
Cloud evidence reviewed 2026-11-16. Full listing and qualifications
Elastic Cloud on Kubernetes
Vector search and licensed machine-learning capabilities support retrieval and AI-oriented workloads.
This entry covers self-managed Elastic Cloud on Kubernetes, not Elastic Cloud Hosted or Elastic Cloud Enterprise. ECK automates application lifecycle but does not transfer cluster, storage, backup or upgrade responsibility to Elastic. The free Basic license and paid Enterprise feature set differ; optional Cloud Connect services create separate external dependencies. ECK 3.5.0 fixes service-token exposure, cross-namespace certificate injection, and retained cross-namespace credentials. CVE-2026-72648 has no workaround; after upgrading, rotate the affected service token as Elastic directs.
- Elastic Cloud on Kubernetes on EKS, AKS and GKE ↗
Vendor / upstream documentation · Reviewed 2026-11-17. Vendor support commitment: not assessed; confirm the selected release and contract.
Elastic documents ECK for managed Kubernetes services including Amazon EKS, Azure Kubernetes Service and Google Kubernetes Engine. The customer owns and operates the cluster and persistent storage.
Cloud evidence reviewed 2026-11-17. Full listing and qualifications
Hyperscience Hypercell
Commercial document-intake and extraction platform installed into customer Kubernetes infrastructure.
The Helm guide supplies complete AWS and GCP values, including application storage permissions. Registry integration alone is not a hosting badge. Hypercell application versions, Helm charts and Kubernetes support dates are separate. The October 2026 decoder advisory and the current v44 flow/block upgrade instructions must be checked before exposing document processing; hosted FedRAMP claims do not transfer to this installation. The vendor Azure reference diagram depicts the complete platform, Blob storage and PostgreSQL; its namespace label still says GKE, so validate the Azure-specific values with the vendor. Apply the fix for the installed Hypercell line and follow the documented sequential upgrades or temporary workarounds.
- Kubernetes Installation Overview ↗
Vendor / upstream documentation · Reviewed 2026-11-19. Vendor support commitment: not assessed; confirm the selected release and contract.
Hyperscience Helm application, database, object storage, workers and separately prepared container registry on AWS or GKE Standard. The Helm guide supplies complete AWS and GCP values, including application storage permissions. Registry integration alone is not a hosting badge. Hypercell application versions, Helm charts and Kubernetes support dates are separate. The October 2026 decoder advisory and the current v44 flow/block upgrade instructions must be checked before exposing document processing; hosted FedRAMP claims do not transfer to this installation. The vendor Azure reference diagram depicts the complete platform, Blob storage and PostgreSQL; its namespace label still says GKE, so validate the Azure-specific values with the vendor.
Cloud evidence reviewed 2026-11-19. Full listing and qualifications
KNIME Business Hub
Provides customer-operated execution and governance for data-science, machine-learning and generative-AI workflows.
This entry covers customer-hosted KNIME Business Hub, not the open-source KNIME Analytics Platform or KNIME's managed Hub offering. Marketplace deployments use single VM images; production capacity, availability and recovery requirements must be assessed separately. Customer-provisioned Kubernetes installation requires the applicable Self Hosted Premium package. Cloud infrastructure charges are separate from KNIME software pricing.
- KNIME Business Hub on Azure Marketplace ↗
Vendor / upstream documentation · Reviewed 2026-11-16. Vendor support commitment: not assessed; confirm the selected release and contract.
KNIME documents launching Business Hub from an Azure Marketplace single-VM image into the customer's Azure subscription.
Cloud evidence reviewed 2026-11-16. Full listing and qualifications
Langflow Self-Hosted
Visual AI workflow and agent builder with Docker and persistent PostgreSQL deployment.
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.
- Langflow Kubernetes deployment architecture ↗
Vendor / upstream documentation · Reviewed 2026-11-12. Vendor support commitment: not assessed; confirm the selected release and contract.
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.
Cloud evidence reviewed 2026-11-12. Full listing and qualifications
Langfuse Self-Hosted
Self-hosted LLM observability, evaluations and prompt-management platform.
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.
- Langfuse customer cloud deployment on Azure ↗
Vendor / upstream documentation · Reviewed 2026-11-12. Vendor support commitment: not assessed; confirm the selected release and contract.
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.
Cloud evidence reviewed 2026-11-12. Full listing and qualifications
LiteLLM Proxy Self-Hosted
Self-hosted AI API gateway providing model routing, access controls and usage tracking.
The gateway forwards each request to its configured model provider. Self-hosting the proxy does not ensure prompts stay inside the account: external model APIs, guardrail services and logging integrations can receive data. Use appropriate self-hosted inference endpoints and explicitly review outbound integrations when required. For release integrity, release images and the Helm chart starting with v1.111.0-dev.3 use keyless Cosign signatures bound to LiteLLM's release workflow; earlier releases require the documented historical public keys. The October 21 incident affected PyPI v1.107.0 and Docker Hub/GHCR v1.107.0 plus mutable latest, main-latest and main-stable tags during the documented exposure windows. Check the incident's artifact digests and remediation instructions; the compromised images carried valid historical signatures, so signature verification alone does not establish safety. LiteLLM v1.110.0 changes authentication, guardrail, accounting, and S3 behavior. The 1.108.11 LTS patch also fixes Bedrock guarded_text handling and restricts os.environ team settings to proxy administrators; review both the selected release line and existing incident-response or image-signing guidance. LiteLLM v1.108.12 LTS fixes tool_policy allowing blocked /v1/messages tool names longer than 64 characters. The standard v1.110.1 UI image changes to slim nginx and removes curl while retaining wget and nc, so retest health probes and scripts that assume curl.
- LiteLLM production deployment ↗
Vendor / upstream documentation · Reviewed 2026-11-12. Vendor support commitment: not assessed; confirm the selected release and contract.
Official production guidance covers EKS, AKS and GKE with Helm, plus Terraform on AWS and Google Cloud. The proxy, database and Redis deployment is separate from the model services it calls.
Cloud evidence reviewed 2026-11-12. Full listing and qualifications
Milvus
Embedding storage and vector retrieval infrastructure for AI search and RAG applications.
Current installation scripts can track a different image than a selected stable release; use versioned configurations for reproducibility. Model inference and embedding generation are separate workloads. Managed Zilliz offerings are outside this listing.
- Milvus deployment on Azure Kubernetes Service ↗
Vendor / upstream documentation · Reviewed 2026-11-12. Vendor support commitment: not assessed; confirm the selected release and contract.
Official instructions provision Azure Kubernetes Service and deploy Milvus using Helm. Review persistent storage, object storage, networking and cluster sizing for the selected version.
Cloud evidence reviewed 2026-11-12. Full listing and qualifications
MLflow Tracking Server Self-Hosted
Self-hosted experiment tracking and LLM/agent observability with customer-controlled metadata and artifacts.
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.
- Self-hosted MLflow deployment on Azure ↗
Vendor / upstream documentation · Reviewed 2026-11-12. Vendor support commitment: not assessed; confirm the selected release and contract.
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.
Cloud evidence reviewed 2026-11-12. Full listing and qualifications
n8n Self-Hosted
Orchestrates AI providers, agents, tools and memory as workflows. Model inference and external service dependencies are configured separately.
This entry excludes n8n Cloud. The Sustainable Use License has use restrictions and should not be described as unrestricted open-source licensing. Enterprise components require the appropriate commercial license. External integrations, AI providers and optional assistant services require separate data-flow review. n8n@2.48.6 and the maintained n8n@1.123.99 line move official images to patched Node runtimes carrying a V8 process-abort fix. The 1.x image also moves to Node 24.22.3, while 2.x pins a patched Node 26.13.0. Update the selected line and verify custom images use the fixed runtime.
- n8n deployment on Azure ↗
Vendor / upstream documentation · Reviewed 2026-11-12. Vendor support commitment: not assessed; confirm the selected release and contract.
Project instructions deploy n8n with PostgreSQL on Azure Kubernetes Service in the customer subscription.
Cloud evidence reviewed 2026-11-12. Full listing and qualifications
NVIDIA AI Enterprise
Provides an enterprise-supported suite of AI frameworks, infrastructure software, operations components and development tools.
This entry covers customer-operated NVIDIA AI Enterprise software, not a managed AI service. A license is required and is not included with most standard, GPU-optimized VMI or managed-Kubernetes deployments; on-demand NVIDIA AI Enterprise VMIs and private offers use separate included-license processes. Not every component is available on every cloud or deployment model. GPU deployments are generally licensed per GPU, while CPU-only instances require an instance license. Tencent Cloud and ByteDance Volcano Engine have documented standard-instance BYOL paths. These guides require certified software and supported instance types; they do not establish that every GPU, component or provider image is eligible. Their per-instance wording differs from the general per-GPU licensing guidance, so confirm the applicable licensing unit. The November lifecycle matrix lists active 9.0/R620, 8.3/R595, and 7.9 LTSB/R580 branches with different end dates. NeMo Microservices support ended October 1, 2026, after 25.12 became the final GA release; select and patch each component branch independently.
- NVIDIA AI Enterprise on AWS, Azure and Google Cloud ↗
Vendor / upstream documentation · Reviewed 2026-11-16. Vendor support commitment: not assessed; confirm the selected release and contract.
NVIDIA documents standard instances, NVIDIA VM images, GPU-optimized VM images, managed Kubernetes and marketplace OpenShift paths across AWS, Azure and Google Cloud. License inclusion varies by image and offer; standard instances and managed Kubernetes generally require BYOL.
Cloud evidence reviewed 2026-11-16. Full listing and qualifications
Onyx Self-Hosted
Provides self-hosted indexing, retrieval, model integration and agent services for building internal AI experiences.
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 on Azure ↗
Vendor / upstream documentation · Reviewed 2026-11-14. Vendor support commitment: not assessed; confirm the selected release and contract.
Onyx documents self-hosting on an Azure VM and an AKS Terraform path using customer Azure managed services.
Cloud evidence reviewed 2026-11-14. Full listing and qualifications
Open WebUI Self-Hosted
Self-hosted AI chat and knowledge interface for local model servers and cloud APIs.
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.
- Open WebUI deployment on Azure Container Apps ↗
Vendor / upstream documentation · Reviewed 2026-11-12. Vendor support commitment: not assessed; confirm the selected release and contract.
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.
Cloud evidence reviewed 2026-11-12. Full listing and qualifications
RAGFlow
Self-hosted document ingestion, retrieval and context infrastructure for AI applications.
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.
- cloudimg RAGFlow Azure VM deployment and operations guide ↗
Third party package · Reviewed 2026-11-12 · Scope: Community / upstream open-source; third-party distribution where noted · Version: 0.26.4. Vendor support commitment: not assessed; confirm the selected release and contract.
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.
Cloud evidence reviewed 2026-11-12. Full listing and qualifications
Red Hat OpenShift AI Self-Managed
AI workbenches, pipelines, model serving and lifecycle tooling delivered on a customer-operated OpenShift platform.
This entry covers OpenShift AI Self-Managed, not Red Hat's managed cloud-service add-on. Support depends on a compatible OpenShift Container Platform release, storage and accelerator matrix, and active subscriptions. Underlying OpenShift and cloud infrastructure are operated by the customer. RHSA-2026:95872 is a critical OpenShift AI 3.5.3 security update covering nine CVEs. Use Red Hat's package and deployment scope to determine affected components, then apply 3.5.3 or a later supported update.
- OpenShift AI Self-Managed on supported OpenShift public-cloud platforms ↗
Vendor / upstream documentation · Reviewed 2026-11-17. Vendor support commitment: not assessed; confirm the selected release and contract.
Red Hat's Self-Managed matrix explicitly lists OpenShift Container Platform on AWS, Azure, Google Cloud, IBM Cloud and Oracle Cloud. The selected OpenShift release, subscriptions, storage, accelerator and component restrictions apply; managed cloud-service add-ons are a separate offering.
Cloud evidence reviewed 2026-11-17. Full listing and qualifications
SAS Viya Platform
Model development, machine learning, model management and AI operationalization; exact products depend on the licensed order.
The public SAS infrastructure repositories are Apache-licensed deployment tooling; SAS Viya itself is commercial software and requires licensed order assets and a JWT license. Kubernetes and PostgreSQL are mandatory. SAS states that its infrastructure-as-code projects satisfy most, not all, deployment requirements. The viya4-deployment project can patch the same manifest but does not move an installation to a different cadence, version or software offering. SAS November bulletins cover compromised eMudhra certificates and critical Brotli CVE-2026-111497 across specified Viya branches. Use the vendor matrices to identify the exact affected component and fixed maintenance level; do not treat a generic Viya version as sufficient.
- SAS Viya infrastructure on Azure ↗
Vendor / upstream documentation · Reviewed 2026-11-17. Vendor support commitment: not assessed; confirm the selected release and contract.
Official SAS Terraform provisions customer Azure resources including AKS, storage and optional PostgreSQL, after which the customer deploys licensed Viya assets.
Cloud evidence reviewed 2026-11-17. Full listing and qualifications
Strategy Enterprise in Your Cloud
Enterprise includes customizable AI agents and AI-assisted analysis capabilities; exact features and model integrations follow the contracted platform.
This entry covers the Strategy Enterprise customer-cloud option, not Strategy Standard in the vendor's managed environment. Strategy describes the service as fully managed despite its deployment in a customer-licensed AWS, Azure Google Cloud or STACKIT environment, so it is classified as BYOC rather than fully customer-operated. Confirm contract-specific account ownership, cloud-resource billing, service availability, support boundaries and cloud-commitment eligibility before procurement.
- Strategy Enterprise in customer-licensed AWS, Azure or Google Cloud ↗
Vendor / upstream documentation · Reviewed 2026-11-17. Vendor support commitment: not assessed; confirm the selected release and contract.
Strategy documents a dedicated single-tenant deployment in a customer-licensed AWS, Azure or GCP environment and states that its cloud team administers and maintains the service on the customer's behalf. The pricing page separately limits own-cloud deployment to Enterprise.
Cloud evidence reviewed 2026-11-17. Full listing and qualifications
vLLM Self-Hosted
Self-hosted model inference and serving engine with OpenAI-compatible APIs.
This entry covers the inference server, not a hosted model service. It can run supported model weights on customer infrastructure, but those weights may need external downloads or access approval and are not licensed by vLLM. Review default usage telemetry and optional outbound integrations; hosting inference does not automatically create an air-gapped system. vLLM v0.34.0 removes Model Runner V1; configurations setting VLLM_USE_V2_MODEL_RUNNER=0 now fail at startup. It changes the default block size from 16 to 64, enables extensible KV cache by default with gpu_memory_utilization defaulting to 1.0, drops ROCm 7.2 wheels and images in favor of ROCm 10.0, requires huggingface_hub 2.0.0 or later, and removes QuTLASS/fp_quant and Marlin W4A8 support. Review the upstream breaking-change table and test model, quantization, memory, and accelerator compatibility before rollout.
- vLLM Production Stack deployment on Azure Kubernetes Service ↗
Vendor / upstream documentation · Reviewed 2026-11-12. Vendor support commitment: not assessed; confirm the selected release and contract.
The project publishes Azure Kubernetes Service provisioning and vLLM deployment scripts in its versioned Production Stack guide. Verify current chart versions, model licenses, accelerator availability, GPU quotas and storage; this is customer-hosted inference, not a hosted model API.
Cloud evidence reviewed 2026-11-12. Full listing and qualifications
Deployment and procurement checks
- Private offers
- Microsoft Marketplace private offers negotiate product pricing and terms. Confirm the exact offer, billing scope and benefit eligibility rather than assuming all purchases count toward a commitment. Provider source ↗ · Reviewed 2026-11-14
- Private connectivity
- Azure Private Link uses private endpoints in a virtual network for supported Azure and partner services. Check application support, DNS and service availability. Provider source ↗ · Reviewed 2026-11-14
- Cross-tenant operations
- Azure Lighthouse supports delegated management with customer-controlled scopes and actions. Agree the roles, audit requirements and revocation plan for the selected vendor. Provider source ↗ · Reviewed 2026-11-14
- Government environments
- Azure Government uses isolated US datacenters and networks. Commercial Azure documentation alone does not establish a product path or compliance in Government. Provider source ↗ · Reviewed 2026-11-14
For every provider, confirm marketplace/private-offer terms, commitment eligibility, regional service availability, private connectivity and any vendor access against your selected deployment. Provider capabilities do not automatically establish application support.
Plan this deployment
Account boundary: Subscription and resource group. Access: Microsoft Entra identities and Azure RBAC. Network: Virtual Network (VNet).
Provider documentation ↗ · Guidance reviewed 2026-11-10. Confirm the application edition, region, support access, backup locations and operating responsibilities before choosing.
Compare this category across clouds · Search these listings · Compare 2–4 listings