swagger: '2.0' info: title: AKASH - gRPC Gateway docs Addresses Gpu API description: A REST interface for state queries version: 1.0.0 tags: - name: Gpu paths: /v1/gpu: get: summary: Get a list of gpu models and their availability. tags: - Gpu security: [] parameters: - schema: type: string required: false name: provider in: query - schema: type: string required: false name: vendor in: query - schema: type: string required: false name: model in: query - schema: type: string required: false name: memory_size in: query responses: '200': description: List of gpu models and their availability. content: application/json: schema: type: object properties: gpus: type: object properties: total: type: object properties: allocatable: type: number allocated: type: number required: - allocatable - allocated details: type: object additionalProperties: type: array items: type: object properties: model: type: string ram: type: string interface: type: string allocatable: type: number allocated: type: number required: - model - ram - interface - allocatable - allocated required: - total - details required: - gpus '400': description: Invalid provider parameter, should be a valid akash address or host uri /v1/gpu-models: get: summary: Get a list of gpu models per vendor. Based on the content from https://raw.githubusercontent.com/akash-network/provider-configs/main/devices/pcie/gpus.json. tags: - Gpu security: [] responses: '200': description: List of gpu models per. content: application/json: schema: type: array items: type: object properties: name: type: string models: type: array items: type: object properties: name: type: string memory: type: array items: type: string interface: type: array items: type: string required: - name - memory - interface required: - name - models /v1/gpu-breakdown: get: tags: - Gpu security: [] summary: Gets gpu analytics breakdown by vendor and model. If no vendor or model is provided, all GPUs are returned. parameters: - schema: type: string required: false name: vendor in: query - schema: type: string required: false name: model in: query responses: '200': description: Gets gpu analytics breakdown by vendor and model. If no vendor or model is provided, all GPUs are returned. content: application/json: schema: type: array items: type: object properties: date: type: string vendor: type: string model: type: string providerCount: type: number nodeCount: type: number totalGpus: type: number leasedGpus: type: number gpuUtilization: type: number required: - date - vendor - model - providerCount - nodeCount - totalGpus - leasedGpus - gpuUtilization /v1/gpu-prices: get: summary: Get a list of gpu models with their availability and pricing. operationId: listGpuPrices tags: - Gpu security: [] responses: '200': description: List of gpu models with their availability and pricing. content: application/json: schema: type: object properties: availability: type: object properties: total: type: number available: type: number required: - total - available models: type: array items: type: object properties: vendor: type: string model: type: string ram: type: string interface: type: string availability: type: object properties: total: type: number available: type: number required: - total - available providerAvailability: type: object properties: total: type: number available: type: number required: - total - available price: type: object nullable: true properties: currency: type: string example: USD min: type: number max: type: number avg: type: number weightedAverage: type: number med: type: number required: - currency - min - max - avg - weightedAverage - med required: - vendor - model - ram - interface - availability - providerAvailability - price required: - availability - models securityDefinitions: kms: type: basic