> ## Documentation Index > Fetch the complete documentation index at: https://docs.together.ai/llms.txt > Use this file to discover all available pages before exploring further. # Create a GPU cluster > Create an Instant Cluster on Together's high-performance GPU clusters. With features like on-demand scaling, long-lived resizable high-bandwidth shared DC-local storage, Kubernetes and Slurm cluster flavors, a REST API, and Terraform support, you can run workloads flexibly without complex infrastructure management. ## OpenAPI ````yaml POST /compute/clusters openapi: 3.1.0 info: title: Together APIs description: The Together REST API. See https://docs.together.ai for more details. version: 2.0.0 termsOfService: https://www.together.ai/terms-of-service contact: name: Together Support url: https://www.together.ai/contact license: name: MIT url: https://github.com/togethercomputer/openapi/blob/main/LICENSE servers: - url: https://api.together.ai/v1 security: - bearerAuth: [] paths: /compute/clusters: post: tags: - GPUClusterService summary: Create a GPU cluster description: > Create an Instant Cluster on Together's high-performance GPU clusters. With features like on-demand scaling, long-lived resizable high-bandwidth shared DC-local storage, Kubernetes and Slurm cluster flavors, a REST API, and Terraform support, you can run workloads flexibly without complex infrastructure management. operationId: GPUClusterService_Create requestBody: content: application/json: schema: $ref: '#/components/schemas/GPUClusterCreateRequest' required: true responses: '200': description: OK content: application/json: schema: $ref: '#/components/schemas/GPUClusterInfo' x-codeSamples: - lang: Python label: Together AI SDK (v2) source: | from together import Together client = Together() response = client.beta.clusters.create( cluster_name="my-gpu-cluster", region="us-central-8", gpu_type="H100_SXM", num_gpus=8, driver_version="CUDA_12_6_560", billint_type="ON_DEMAND", ) print(response.cluster_id) - lang: TypeScript label: Together AI SDK (v2) source: | import Together from "together-ai"; const client = new Together(); const response = await client.beta.clusters.create({ cluster_name: "my-gpu-cluster", region: "us-central-8", gpu_type: "H100_SXM", num_gpus: 8, driver_version: "CUDA_12_6_560", billint_type: "ON_DEMAND", }); console.log(response.cluster_id) - lang: JavaScript label: Together AI SDK (v2) source: | import Together from "together-ai"; const client = new Together(); const response = await client.beta.clusters.create({ cluster_name: "my-gpu-cluster", region: "us-central-8", gpu_type: "H100_SXM", num_gpus: 8, driver_version: "CUDA_12_6_560", billint_type: "ON_DEMAND", }); console.log(response.cluster_id) - lang: Shell label: CLI source: | curl -X POST \ -H "Authorization Bearer $TOGETHER_API_KEY" \ --data '{ "region": "us-west-2", "gpu_type": "H100_SXM", "num_gpus": 8, "cluster_name": "my-gpu-cluster", "driver_version": "CUDA_12_6_560" }' \ https://api.together.ai/v1/compute/clusters components: schemas: GPUClusterCreateRequest: description: GPU Cluster create request required: - region - gpu_type - num_gpus - cluster_name - cuda_version - nvidia_driver_version - billing_type type: object properties: cluster_type: description: Type of cluster to create. type: string enum: - KUBERNETES - SLURM region: description: >- Region to create the GPU cluster in. Usable regions can be found from `client.clusters.list_regions()` type: string gpu_type: description: Type of GPU to use in the cluster type: string enum: - H100_SXM - H200_SXM - RTX_6000_PCI - L40_PCIE - B200_SXM - H100_SXM_INF num_gpus: description: >- Number of GPUs to allocate in the cluster. This must be multiple of 8. For example, 8, 16 or 24 type: integer cluster_name: description: Name of the GPU cluster. type: string duration_days: x-stainless-terraform-configurability: computed description: Duration in days to keep the cluster running. type: integer shared_volume: $ref: '#/components/schemas/GPUClustersSharedVolumeCreateRequest' x-stainless-terraform-configurability: computed description: >- Inline configuration to create a shared volume with the cluster creation. volume_id: description: ID of an existing volume to use with the cluster creation. type: string billing_type: description: > RESERVED billing types allow you to specify the duration of the cluster reservation via the duration_days field. ON_DEMAND billing types will give you ownership of the cluster until you delete it. SCHEDULED_CAPACITY billing types allow you to reserve capacity for a scheduled time window. You must specify the reservation_start_time and reservation_end_time with this request. x-stainless-terraform-configurability: computed type: string enum: - RESERVED - ON_DEMAND - SCHEDULED_CAPACITY gpu_node_failover_enabled: type: boolean default: false description: >- Whether automated GPU node failover should be enabled for this cluster. By default, it is disabled. auto_scaled: type: boolean default: false description: >- Whether GPU cluster should be auto-scaled based on the workload. By default, it is not auto-scaled. auto_scale_max_gpus: type: integer description: >- Maximum number of GPUs to which the cluster can be auto-scaled up. This field is required if auto_scaled is true. slurm_shm_size_gib: type: integer description: >- Shared memory size in GiB for Slurm cluster. This field is required if cluster_type is SLURM. capacity_pool_id: type: string description: >- ID of the capacity pool to use for the cluster. This field is optional and only applicable if the cluster is created from a capacity pool. reservation_start_time: type: string description: >- Reservation start time of the cluster. This field is required for SCHEDULED billing to specify the reservation start time for the cluster. If not provided, the cluster provisions immediately. format: date-time reservation_end_time: type: string description: >- Reservation end time of the cluster. This field is required for SCHEDULED billing to specify the reservation end time for the cluster. format: date-time install_traefik: type: boolean default: false description: >- Whether to install Traefik ingress controller in the cluster. This field is only applicable for Kubernetes clusters and is false by default. cuda_version: type: string description: CUDA version for this cluster. For example, 12.5 nvidia_driver_version: type: string description: >- Nvidia driver version for this cluster. For example, 550. Only some combination of cuda_version and nvidia_driver_version are supported. slurm_image: type: string description: Custom Slurm image for Slurm clusters. GPUClusterInfo: type: object required: - cluster_id - cluster_type - region - gpu_type - cluster_name - cuda_version - nvidia_driver_version - volumes - status - control_plane_nodes - gpu_worker_nodes - kube_config - num_gpus properties: cluster_id: type: string cluster_type: description: Type of cluster. enum: - KUBERNETES - SLURM region: type: string gpu_type: enum: - H100_SXM - H200_SXM - RTX_6000_PCI - L40_PCIE - B200_SXM - H100_SXM_INF cluster_name: type: string duration_hours: type: integer volumes: type: array items: $ref: '#/components/schemas/GPUClusterVolume' status: description: Current status of the GPU cluster. enum: - WaitingForControlPlaneNodes - WaitingForDataPlaneNodes - WaitingForSubnet - WaitingForSharedVolume - InstallingDrivers - RunningAcceptanceTests - Paused - OnDemandComputePaused - Ready - Degraded - Deleting control_plane_nodes: type: array items: $ref: '#/components/schemas/GPUClusterControlPlaneNode' gpu_worker_nodes: type: array items: $ref: '#/components/schemas/GPUClusterGPUWorkerNode' kube_config: type: string num_gpus: type: integer slurm_shm_size_gib: type: integer capacity_pool_id: type: string reservation_start_time: type: string format: date-time reservation_end_time: type: string format: date-time install_traefik: type: boolean cuda_version: type: string nvidia_driver_version: type: string created_at: type: string format: date-time GPUClustersSharedVolumeCreateRequest: type: object required: - volume_name - size_tib - region properties: volume_name: description: Customizable name of the volume to create. type: string size_tib: description: Volume size in whole tebibytes (TiB). type: integer region: type: string description: >- Region name. Usable regions can be found from `client.clusters.list_regions()` GPUClusterVolume: type: object required: - volume_id - volume_name - size_tib - status properties: volume_id: type: string volume_name: type: string size_tib: type: integer status: type: string GPUClusterControlPlaneNode: type: object required: - node_id - node_name - status - host_name - num_cpu_cores - memory_gib - network properties: node_id: type: string node_name: type: string status: type: string host_name: type: string num_cpu_cores: type: integer memory_gib: type: number network: type: string GPUClusterGPUWorkerNode: type: object required: - node_id - node_name - status - host_name - num_cpu_cores - num_gpus - memory_gib - networks properties: node_id: type: string node_name: type: string status: type: string host_name: type: string num_cpu_cores: type: integer num_gpus: type: integer memory_gib: type: number networks: type: array items: type: string instance_id: type: string securitySchemes: bearerAuth: type: http scheme: bearer x-bearer-format: bearer x-default: default ````