# Gcore AI > Deploy AI inference and GPU-accelerated training using Gcore AI products - Everywhere Inference for low-latency model serving at the edge and GPU Cloud for Bare Metal and Virtual Machine compute. - [Overview](https://docs.gcore.com/edge-ai.md): Deploy AI inference and GPU-accelerated training using Gcore AI products - Everywhere Inference for low-latency model serving at the edge and GPU Cloud for Bare Metal and Virtual Machine compute. - [Getting started](https://docs.gcore.com/edge-ai/getting-started.md): Deploy AI models for inference using Gcore Everywhere Inference via Customer Portal or API, or create GPU clusters for model training with Gcore GPU Cloud and bare-metal infrastructure. - [Billing](https://docs.gcore.com/edge-ai/billing.md): Configure billing for Gcore GPU Virtual Machines and Bare Metal instances charged by resource type and configuration, and Everywhere Inference resources billed per minute of compute time with monthly fee processing and service suspension on payment failure. ## GPU Cloud - [About GPU Cloud](https://docs.gcore.com/edge-ai/ai-infrastructure/about-gpu-cloud.md): Deploy GPU clusters with Bare Metal GPU, Spot Bare Metal GPU, or Virtual GPU instance types, configured with NVIDIA H200, H100, A100, or L40S GPUs, InfiniBand or Ethernet interconnect, and NVMe storage for distributed LLM training, inference, and HPC workloads. - [Create a Virtual GPU cluster](https://docs.gcore.com/edge-ai/ai-infrastructure/create-a-virtual-gpu-cluster.md): Create Virtual GPU clusters with configurable node count, GPU models, flavors, volumes, and network settings via Customer Portal, REST API, or Terraform. - [Manage a Virtual GPU cluster](https://docs.gcore.com/edge-ai/ai-infrastructure/manage-a-virtual-gpu-cluster.md): Manage Virtual GPU clusters via Customer Portal, REST API, or Terraform; rename, scale, control power state, manage volumes and interfaces, and delete clusters. - [Create a Bare Metal GPU cluster](https://docs.gcore.com/edge-ai/ai-infrastructure/create-a-bare-metal-gpu-cluster.md): Create Bare Metal GPU clusters with dedicated physical servers via Customer Portal, REST API, or Terraform. - [Manage a Bare Metal GPU cluster](https://docs.gcore.com/edge-ai/ai-infrastructure/manage-a-bare-metal-gpu-cluster.md): Manage Bare Metal GPU cluster nodes, resize, control power states, reconfigure cluster specifications, apply settings to outdated nodes, rebuild, replace, and delete nodes via Customer Portal, REST API, or Terraform. - [Spot Bare Metal GPU](https://docs.gcore.com/edge-ai/ai-infrastructure/spot-bare-metal-gpu.md): Deploy Spot Bare Metal GPU clusters with discounted pricing for interruptible workloads via Customer Portal or REST API - check availability, create clusters when capacity allows, and understand reclamation with 24-hour notice. ## Managed Kubernetes - [Overview](https://docs.gcore.com/edge-ai/managed-kubernetes/overview.md) - [Create a GPU Kubernetes cluster](https://docs.gcore.com/edge-ai/managed-kubernetes/create-a-gpu-kubernetes-cluster.md): Create GPU Kubernetes clusters with Bare Metal GPU pools via Customer Portal, REST API, or Terraform - configure version, pools, network, and download kubeconfig. ## Manage a cluster - [Connect to a GPU Kubernetes cluster](https://docs.gcore.com/edge-ai/managed-kubernetes/connect-to-a-gpu-kubernetes-cluster.md): Connect to a Gcore GPU Kubernetes cluster via Customer Portal, REST API, or Terraform - download kubeconfig, configure kubectl, and list or inspect clusters. - [Manage node pools](https://docs.gcore.com/edge-ai/managed-kubernetes/manage-node-pools.md): Manage GPU Kubernetes node pools via Customer Portal, REST API, or Terraform - view, scale, add, replace nodes, and delete pools. - [Configure cluster autoscaling](https://docs.gcore.com/edge-ai/managed-kubernetes/configure-cluster-autoscaling.md): Configure the Cluster Autoscaler for a Gcore GPU Kubernetes cluster via the Customer Portal, REST API, or Terraform. - [Configure cluster logging](https://docs.gcore.com/edge-ai/managed-kubernetes/configure-cluster-logging.md): Configure container and system log collection for a Gcore GPU Kubernetes cluster via the Customer Portal, REST API, or Terraform. - [Configure OIDC authentication](https://docs.gcore.com/edge-ai/managed-kubernetes/configure-oidc-authentication.md): Configure OIDC Single Sign-On for Kubernetes API access in a Gcore GPU Kubernetes cluster via the Customer Portal, REST API, or Terraform. - [Upgrade a GPU Kubernetes cluster](https://docs.gcore.com/edge-ai/managed-kubernetes/upgrade-a-gpu-kubernetes-cluster.md): Update a Gcore GPU Kubernetes cluster to the next minor version via Customer Portal, REST API, or Terraform. - [View cluster audit log](https://docs.gcore.com/edge-ai/managed-kubernetes/view-cluster-audit-log.md): View the audit log of GPU Kubernetes cluster operations via the Gcore Customer Portal. - [Delete a GPU Kubernetes cluster](https://docs.gcore.com/edge-ai/managed-kubernetes/delete-a-gpu-kubernetes-cluster.md): Delete a Gcore GPU Kubernetes cluster via Customer Portal, REST API, or Terraform. ## Everywhere Inference - [Overview](https://docs.gcore.com/edge-ai/everywhere-inference.md): Deploy trained AI models on Gcore Everywhere Inference edge nodes using anycast endpoints and smart routing to achieve low-latency inference across global regions with automatic failover. - [Application Catalog](https://docs.gcore.com/edge-ai/everywhere-inference/application-catalog.md): Browse and deploy pre-built open-source AI models from the Gcore Everywhere Inference Application Catalog, including text LLMs, image generation, speech recognition, text embedding, and other model categories. ## AI models - [Deploy a model from the Application Catalog](https://docs.gcore.com/edge-ai/everywhere-inference/ai-models/deploy-from-catalog.md): Deploy a pre-built open-source AI model from the Gcore Everywhere Inference Application Catalog by selecting an application, choosing compute resources and regions, and configuring deployment settings in the Create Deployment form. - [Prepare a custom AI model for deployment](https://docs.gcore.com/edge-ai/everywhere-inference/ai-models/prepare-a-custom-ai-model-for-deployment.md): Prepare custom AI models for deployment by containerizing with vLLM or Diffusers, optimizing via quantization and pruning, building Docker images from Hugging Face models, and pushing to a container registry for inference serving. - [Deploy a custom model](https://docs.gcore.com/edge-ai/everywhere-inference/ai-models/deploy-a-custom-model.md): Deploy a custom AI model to Gcore Everywhere Inference using a Docker container image from a public or private registry, with configurable compute resources, routing regions, autoscaling, health checks, and optional API key authentication. - [Manage AI model deployments](https://docs.gcore.com/edge-ai/everywhere-inference/ai-models/manage-ai-model-deployments.md): Manage AI model deployments in Everywhere Inference by viewing deployment status, endpoint URLs, replica counts, and logs; navigate to deployment details to modify settings such as routing placement, flavor, and pod counts; delete deployments from the action menu. - [Query a deployed model](https://docs.gcore.com/edge-ai/everywhere-inference/ai-models/query-deployed-model.md): Send inference requests to a deployed Everywhere Inference model using curl, Python OpenAI SDK, or JavaScript. Covers finding the endpoint URL, forming chat completion requests, multi-turn conversations, streaming, and key request parameters. - [Chat with a model via Open WebUI](https://docs.gcore.com/edge-ai/everywhere-inference/ai-models/chat-with-open-webui.md): Deploy a Gcore Everywhere Inference model with Open WebUI enabled and interact with it through a browser-based chat interface. Covers enabling the Open WebUI module during deployment, accessing the chat UI, creating an admin account, and starting a conversation. - [Translate text with Llama 3.2](https://docs.gcore.com/edge-ai/everywhere-inference/ai-models/translate-text-with-llama.md): Use a deployed Llama 3.2 model on Gcore Everywhere Inference to translate text between languages. Covers the system prompt pattern, request parameters, curl, Python, and JavaScript code examples, edge case handling, and output style control. ## Container image registries - [Add a container image registry](https://docs.gcore.com/edge-ai/everywhere-inference/container-image-registries/add-a-registry.md): Add a private container image registry to Gcore Everywhere Inference. Covers what a registry is, how credentials work, registry URLs for Docker Hub, Hugging Face, and GitHub Container Registry, and how the registry connects to a deployment. - [Manage container image registries](https://docs.gcore.com/edge-ai/everywhere-inference/container-image-registries/manage-registries.md): View, edit, and delete container image registries in Gcore Everywhere Inference. Covers how registries are used in deployments, editing authentication credentials, and deleting a registry. ## API keys - [Inference deployment with API key authentication](https://docs.gcore.com/edge-ai/everywhere-inference/api-keys/create-inference-deployment-with-auth.md): Set up API key authentication for an Everywhere Inference deployment by creating a project-level API key, linking it to a deployment, and including the key in the X-API-Key header of every inference request. ## Quotas - [Request a quota increase](https://docs.gcore.com/edge-ai/everywhere-inference/quotas/request-quota-increase.md): Request a quota increase for Gcore Everywhere Inference by selecting GPU or CPU resources in the Quotas Viewer, entering a use case description, and submitting the request. Track request status in Quotas Request History. - [Quotas Request History](https://docs.gcore.com/edge-ai/everywhere-inference/quotas/quotas-request-history.md): View the history of quota increase requests for Gcore Everywhere Inference, including request status, resource type, and requested amounts, in the Quotas Request History page.