@dhbw-cloud/llm-client 0.2.16 → 0.2.17

package/dist/index.d.ts

@@ -1,2 +1,2 @@
-export { blockMachine, createKey, deleteFleetPackage, deleteGpuPool, deleteKey, deleteTreeProject, deleteWorkspaceProject, forgetMachine, getFleet, getFleetInference, getFleetInventoryCsv, getFleetPackage, getFleetProfile, getFleetReadme, getGpuFleet, getGpuHours, getMe, getMyWorkspaceUsage, getProjectImageLog, getTreeCatalog, getUsage, getWorkspaceCatalog, getWorkspaceStorage, getWorkspaceUsage, listGpuServers, listKeys, listTreeProjects, listWorkspaceProjects, type Options, putGpuNode, putGpuPool, putTreeProject, rebuildProjectImage, setDefaultProject, stopGpuServer, unblockMachine, updateWorkspaceProject } from './sdk.gen';
-export type { AccessRole, BlockMachineData, BlockMachineError, BlockMachineErrors, BlockMachineResponse, BlockMachineResponses, ClientOptions, CreateKeyData, CreateKeyError, CreateKeyErrors, CreateKeyResponse, CreateKeyResponses, DeleteFleetPackageData, DeleteFleetPackageResponse, DeleteFleetPackageResponses, DeleteGpuPoolData, DeleteGpuPoolError, DeleteGpuPoolErrors, DeleteGpuPoolResponse, DeleteGpuPoolResponses, DeleteKeyData, DeleteKeyError, DeleteKeyErrors, DeleteKeyResponse, DeleteKeyResponses, DeleteTreeProjectData, DeleteTreeProjectError, DeleteTreeProjectErrors, DeleteTreeProjectResponse, DeleteTreeProjectResponses, DeleteWorkspaceProjectData, DeleteWorkspaceProjectError, DeleteWorkspaceProjectErrors, DeleteWorkspaceProjectResponse, DeleteWorkspaceProjectResponses, FleetFleetView, FleetPackageInfo, FleetPeerView, FleetProfile, ForgetMachineData, ForgetMachineError, ForgetMachineErrors, ForgetMachineResponse, ForgetMachineResponses, GetFleetData, GetFleetError, GetFleetErrors, GetFleetInferenceData, GetFleetInferenceError, GetFleetInferenceErrors, GetFleetInferenceResponse, GetFleetInferenceResponses, GetFleetInventoryCsvData, GetFleetInventoryCsvResponse, GetFleetInventoryCsvResponses, GetFleetPackageData, GetFleetPackageError, GetFleetPackageErrors, GetFleetPackageResponse, GetFleetPackageResponses, GetFleetProfileData, GetFleetProfileResponse, GetFleetProfileResponses, GetFleetReadmeData, GetFleetReadmeResponse, GetFleetReadmeResponses, GetFleetResponse, GetFleetResponses, GetGpuFleetData, GetGpuFleetError, GetGpuFleetErrors, GetGpuFleetResponse, GetGpuFleetResponses, GetGpuHoursData, GetGpuHoursError, GetGpuHoursErrors, GetGpuHoursResponse, GetGpuHoursResponses, GetMeData, GetMeError, GetMeErrors, GetMeResponse, GetMeResponses, GetMyWorkspaceUsageData, GetMyWorkspaceUsageError, GetMyWorkspaceUsageErrors, GetMyWorkspaceUsageResponse, GetMyWorkspaceUsageResponses, GetProjectImageLogData, GetProjectImageLogError, GetProjectImageLogErrors, GetProjectImageLogResponse, GetProjectImageLogResponses, GetTreeCatalogData, GetTreeCatalogError, GetTreeCatalogErrors, GetTreeCatalogResponse, GetTreeCatalogResponses, GetUsageData, GetUsageError, GetUsageErrors, GetUsageResponse, GetUsageResponses, GetWorkspaceCatalogData, GetWorkspaceCatalogError, GetWorkspaceCatalogErrors, GetWorkspaceCatalogResponse, GetWorkspaceCatalogResponses, GetWorkspaceStorageData, GetWorkspaceStorageError, GetWorkspaceStorageErrors, GetWorkspaceStorageResponse, GetWorkspaceStorageResponses, GetWorkspaceUsageData, GetWorkspaceUsageError, GetWorkspaceUsageErrors, GetWorkspaceUsageResponse, GetWorkspaceUsageResponses, GpuClassInfo, GpuGpuNode, GpuInferencePod, GpuInferencePool, GpuNodeAssignment, GpuPool, GpuServer, GpuTier, KeysKeyView, KeysProjectUsage, KeysUsage, ListGpuServersData, ListGpuServersError, ListGpuServersErrors, ListGpuServersResponse, ListGpuServersResponses, ListKeysData, ListKeysError, ListKeysErrors, ListKeysResponse, ListKeysResponses, ListTreeProjectsData, ListTreeProjectsError, ListTreeProjectsErrors, ListTreeProjectsResponse, ListTreeProjectsResponses, ListWorkspaceProjectsData, ListWorkspaceProjectsResponse, ListWorkspaceProjectsResponses, PutGpuNodeData, PutGpuNodeError, PutGpuNodeErrors, PutGpuNodeResponse, PutGpuNodeResponses, PutGpuPoolData, PutGpuPoolError, PutGpuPoolErrors, PutGpuPoolResponse, PutGpuPoolResponses, PutTreeProjectData, PutTreeProjectError, PutTreeProjectErrors, PutTreeProjectResponse, PutTreeProjectResponses, RebuildProjectImageData, RebuildProjectImageError, RebuildProjectImageErrors, RebuildProjectImageResponse, RebuildProjectImageResponses, SetDefaultProjectData, SetDefaultProjectError, SetDefaultProjectErrors, SetDefaultProjectResponse, SetDefaultProjectResponses, StopGpuServerData, StopGpuServerError, StopGpuServerErrors, StopGpuServerResponse, StopGpuServerResponses, UnblockMachineData, UnblockMachineError, UnblockMachineErrors, UnblockMachineResponse, UnblockMachineResponses, UpdateWorkspaceProjectData, UpdateWorkspaceProjectError, UpdateWorkspaceProjectErrors, UpdateWorkspaceProjectResponse, UpdateWorkspaceProjectResponses, UsageGpuHour, UsageGpuHourPerson, UsageGpuReport, UsageReport, UsageTotal, WebserverBuildLogResponse, WebserverCreateKeyRequest, WebserverCreateKeyResponse, WebserverDefaultProjectRequest, WebserverDefaultProjectResponse, WebserverErrorResponse, WebserverGpuFleetResponse, WebserverGpuNodeRequest, WebserverGpuPoolRequest, WebserverMeResponse, WebserverTreeCatalog, WorkspaceBase, WorkspaceBudget, WorkspaceBudgetStatus, WorkspaceConfig, WorkspaceGpuClass, WorkspaceImageBuild, WorkspaceLimits, WorkspaceMember, WorkspaceMine, WorkspaceProfile, WorkspaceProject, WorkspaceStorageDir, WorkspaceStorageReport, WorkspaceTreeBudget, WorkspaceTreeProject, WorkspaceTreeStatus, WorkspaceTreeUse, WorkspaceValkeyMemory } from './types.gen';
+export { blockMachine, createKey, deleteCatalogModel, deleteFleetPackage, deleteGpuPool, deleteKey, deleteTreeProject, deleteWorkspaceProject, forgetMachine, getFleet, getFleetInference, getFleetInventoryCsv, getFleetPackage, getFleetProfile, getFleetReadme, getGpuFleet, getGpuHours, getMe, getModelCatalog, getMyWorkspaceUsage, getProjectImageLog, getTreeCatalog, getUsage, getWorkspaceCatalog, getWorkspaceStorage, getWorkspaceUsage, listGpuServers, listKeys, listTreeProjects, listWorkspaceProjects, type Options, putCatalogModel, putGpuNode, putGpuPool, putTierModel, putTreeProject, rebuildProjectImage, setDefaultProject, stopGpuServer, unblockMachine, updateWorkspaceProject } from './sdk.gen';
+export type { AccessRole, BlockMachineData, BlockMachineError, BlockMachineErrors, BlockMachineResponse, BlockMachineResponses, ClientOptions, CreateKeyData, CreateKeyError, CreateKeyErrors, CreateKeyResponse, CreateKeyResponses, DeleteCatalogModelData, DeleteCatalogModelError, DeleteCatalogModelErrors, DeleteCatalogModelResponse, DeleteCatalogModelResponses, DeleteFleetPackageData, DeleteFleetPackageResponse, DeleteFleetPackageResponses, DeleteGpuPoolData, DeleteGpuPoolError, DeleteGpuPoolErrors, DeleteGpuPoolResponse, DeleteGpuPoolResponses, DeleteKeyData, DeleteKeyError, DeleteKeyErrors, DeleteKeyResponse, DeleteKeyResponses, DeleteTreeProjectData, DeleteTreeProjectError, DeleteTreeProjectErrors, DeleteTreeProjectResponse, DeleteTreeProjectResponses, DeleteWorkspaceProjectData, DeleteWorkspaceProjectError, DeleteWorkspaceProjectErrors, DeleteWorkspaceProjectResponse, DeleteWorkspaceProjectResponses, FleetFleetView, FleetPackageInfo, FleetPeerView, FleetProfile, ForgetMachineData, ForgetMachineError, ForgetMachineErrors, ForgetMachineResponse, ForgetMachineResponses, GetFleetData, GetFleetError, GetFleetErrors, GetFleetInferenceData, GetFleetInferenceError, GetFleetInferenceErrors, GetFleetInferenceResponse, GetFleetInferenceResponses, GetFleetInventoryCsvData, GetFleetInventoryCsvResponse, GetFleetInventoryCsvResponses, GetFleetPackageData, GetFleetPackageError, GetFleetPackageErrors, GetFleetPackageResponse, GetFleetPackageResponses, GetFleetProfileData, GetFleetProfileResponse, GetFleetProfileResponses, GetFleetReadmeData, GetFleetReadmeResponse, GetFleetReadmeResponses, GetFleetResponse, GetFleetResponses, GetGpuFleetData, GetGpuFleetError, GetGpuFleetErrors, GetGpuFleetResponse, GetGpuFleetResponses, GetGpuHoursData, GetGpuHoursError, GetGpuHoursErrors, GetGpuHoursResponse, GetGpuHoursResponses, GetMeData, GetMeError, GetMeErrors, GetMeResponse, GetMeResponses, GetModelCatalogData, GetModelCatalogError, GetModelCatalogErrors, GetModelCatalogResponse, GetModelCatalogResponses, GetMyWorkspaceUsageData, GetMyWorkspaceUsageError, GetMyWorkspaceUsageErrors, GetMyWorkspaceUsageResponse, GetMyWorkspaceUsageResponses, GetProjectImageLogData, GetProjectImageLogError, GetProjectImageLogErrors, GetProjectImageLogResponse, GetProjectImageLogResponses, GetTreeCatalogData, GetTreeCatalogError, GetTreeCatalogErrors, GetTreeCatalogResponse, GetTreeCatalogResponses, GetUsageData, GetUsageError, GetUsageErrors, GetUsageResponse, GetUsageResponses, GetWorkspaceCatalogData, GetWorkspaceCatalogError, GetWorkspaceCatalogErrors, GetWorkspaceCatalogResponse, GetWorkspaceCatalogResponses, GetWorkspaceStorageData, GetWorkspaceStorageError, GetWorkspaceStorageErrors, GetWorkspaceStorageResponse, GetWorkspaceStorageResponses, GetWorkspaceUsageData, GetWorkspaceUsageError, GetWorkspaceUsageErrors, GetWorkspaceUsageResponse, GetWorkspaceUsageResponses, GpuCatalogModel, GpuClassInfo, GpuGpuNode, GpuInferencePod, GpuInferencePool, GpuModelCatalog, GpuModelChoice, GpuModelState, GpuNodeAssignment, GpuPool, GpuServer, GpuTier, KeysKeyView, KeysProjectUsage, KeysUsage, ListGpuServersData, ListGpuServersError, ListGpuServersErrors, ListGpuServersResponse, ListGpuServersResponses, ListKeysData, ListKeysError, ListKeysErrors, ListKeysResponse, ListKeysResponses, ListTreeProjectsData, ListTreeProjectsError, ListTreeProjectsErrors, ListTreeProjectsResponse, ListTreeProjectsResponses, ListWorkspaceProjectsData, ListWorkspaceProjectsResponse, ListWorkspaceProjectsResponses, PutCatalogModelData, PutCatalogModelError, PutCatalogModelErrors, PutCatalogModelResponse, PutCatalogModelResponses, PutGpuNodeData, PutGpuNodeError, PutGpuNodeErrors, PutGpuNodeResponse, PutGpuNodeResponses, PutGpuPoolData, PutGpuPoolError, PutGpuPoolErrors, PutGpuPoolResponse, PutGpuPoolResponses, PutTierModelData, PutTierModelError, PutTierModelErrors, PutTierModelResponse, PutTierModelResponses, PutTreeProjectData, PutTreeProjectError, PutTreeProjectErrors, PutTreeProjectResponse, PutTreeProjectResponses, RebuildProjectImageData, RebuildProjectImageError, RebuildProjectImageErrors, RebuildProjectImageResponse, RebuildProjectImageResponses, SetDefaultProjectData, SetDefaultProjectError, SetDefaultProjectErrors, SetDefaultProjectResponse, SetDefaultProjectResponses, StopGpuServerData, StopGpuServerError, StopGpuServerErrors, StopGpuServerResponse, StopGpuServerResponses, UnblockMachineData, UnblockMachineError, UnblockMachineErrors, UnblockMachineResponse, UnblockMachineResponses, UpdateWorkspaceProjectData, UpdateWorkspaceProjectError, UpdateWorkspaceProjectErrors, UpdateWorkspaceProjectResponse, UpdateWorkspaceProjectResponses, UsageGpuHour, UsageGpuHourPerson, UsageGpuReport, UsageReport, UsageTotal, WebserverBuildLogResponse, WebserverCreateKeyRequest, WebserverCreateKeyResponse, WebserverDefaultProjectRequest, WebserverDefaultProjectResponse, WebserverErrorResponse, WebserverGpuFleetResponse, WebserverGpuNodeRequest, WebserverGpuPoolRequest, WebserverMeResponse, WebserverTierModelRequest, WebserverTreeCatalog, WorkspaceBase, WorkspaceBudget, WorkspaceBudgetStatus, WorkspaceConfig, WorkspaceGpuClass, WorkspaceImageBuild, WorkspaceLimits, WorkspaceMember, WorkspaceMine, WorkspaceProfile, WorkspaceProject, WorkspaceStorageDir, WorkspaceStorageReport, WorkspaceTreeBudget, WorkspaceTreeProject, WorkspaceTreeStatus, WorkspaceTreeUse, WorkspaceValkeyMemory } from './types.gen';

package/dist/index.mjs

@@ -875,6 +875,34 @@ var stopGpuServer = (options) => (options.client ?? client).delete({
   url: "/v1/gpu/servers/{name}",
   ...options
 });
+var getModelCatalog = (options) => (options?.client ?? client).get({
+  security: [{ name: "Authorization", type: "apiKey" }],
+  url: "/v1/inference/models",
+  ...options
+});
+var deleteCatalogModel = (options) => (options.client ?? client).delete({
+  security: [{ name: "Authorization", type: "apiKey" }],
+  url: "/v1/inference/models/{key}",
+  ...options
+});
+var putCatalogModel = (options) => (options.client ?? client).put({
+  security: [{ name: "Authorization", type: "apiKey" }],
+  url: "/v1/inference/models/{key}",
+  ...options,
+  headers: {
+    "Content-Type": "application/json",
+    ...options.headers
+  }
+});
+var putTierModel = (options) => (options.client ?? client).put({
+  security: [{ name: "Authorization", type: "apiKey" }],
+  url: "/v1/inference/tiers/{tier}",
+  ...options,
+  headers: {
+    "Content-Type": "application/json",
+    ...options.headers
+  }
+});
 var listKeys = (options) => (options?.client ?? client).get({
   security: [{ name: "Authorization", type: "apiKey" }],
   url: "/v1/keys",
@@ -995,6 +1023,7 @@ export {
   blockMachine,
   client,
   createKey,
+  deleteCatalogModel,
   deleteFleetPackage,
   deleteGpuPool,
   deleteKey,
@@ -1010,6 +1039,7 @@ export {
   getGpuFleet,
   getGpuHours,
   getMe,
+  getModelCatalog,
   getMyWorkspaceUsage,
   getProjectImageLog,
   getTreeCatalog,
@@ -1021,8 +1051,10 @@ export {
   listKeys,
   listTreeProjects,
   listWorkspaceProjects,
+  putCatalogModel,
   putGpuNode,
   putGpuPool,
+  putTierModel,
   putTreeProject,
   rebuildProjectImage,
   setDefaultProject,

package/dist/sdk.gen.d.ts

@@ -1,5 +1,5 @@
 import type { Client, ClientMeta, Options as Options2, RequestResult, TDataShape } from './client';
-import type { BlockMachineData, BlockMachineErrors, BlockMachineResponses, CreateKeyData, CreateKeyErrors, CreateKeyResponses, DeleteFleetPackageData, DeleteFleetPackageResponses, DeleteGpuPoolData, DeleteGpuPoolErrors, DeleteGpuPoolResponses, DeleteKeyData, DeleteKeyErrors, DeleteKeyResponses, DeleteTreeProjectData, DeleteTreeProjectErrors, DeleteTreeProjectResponses, DeleteWorkspaceProjectData, DeleteWorkspaceProjectErrors, DeleteWorkspaceProjectResponses, ForgetMachineData, ForgetMachineErrors, ForgetMachineResponses, GetFleetData, GetFleetErrors, GetFleetInferenceData, GetFleetInferenceErrors, GetFleetInferenceResponses, GetFleetInventoryCsvData, GetFleetInventoryCsvResponses, GetFleetPackageData, GetFleetPackageErrors, GetFleetPackageResponses, GetFleetProfileData, GetFleetProfileResponses, GetFleetReadmeData, GetFleetReadmeResponses, GetFleetResponses, GetGpuFleetData, GetGpuFleetErrors, GetGpuFleetResponses, GetGpuHoursData, GetGpuHoursErrors, GetGpuHoursResponses, GetMeData, GetMeErrors, GetMeResponses, GetMyWorkspaceUsageData, GetMyWorkspaceUsageErrors, GetMyWorkspaceUsageResponses, GetProjectImageLogData, GetProjectImageLogErrors, GetProjectImageLogResponses, GetTreeCatalogData, GetTreeCatalogErrors, GetTreeCatalogResponses, GetUsageData, GetUsageErrors, GetUsageResponses, GetWorkspaceCatalogData, GetWorkspaceCatalogErrors, GetWorkspaceCatalogResponses, GetWorkspaceStorageData, GetWorkspaceStorageErrors, GetWorkspaceStorageResponses, GetWorkspaceUsageData, GetWorkspaceUsageErrors, GetWorkspaceUsageResponses, ListGpuServersData, ListGpuServersErrors, ListGpuServersResponses, ListKeysData, ListKeysErrors, ListKeysResponses, ListTreeProjectsData, ListTreeProjectsErrors, ListTreeProjectsResponses, ListWorkspaceProjectsData, ListWorkspaceProjectsResponses, PutGpuNodeData, PutGpuNodeErrors, PutGpuNodeResponses, PutGpuPoolData, PutGpuPoolErrors, PutGpuPoolResponses, PutTreeProjectData, PutTreeProjectErrors, PutTreeProjectResponses, RebuildProjectImageData, RebuildProjectImageErrors, RebuildProjectImageResponses, SetDefaultProjectData, SetDefaultProjectErrors, SetDefaultProjectResponses, StopGpuServerData, StopGpuServerErrors, StopGpuServerResponses, UnblockMachineData, UnblockMachineErrors, UnblockMachineResponses, UpdateWorkspaceProjectData, UpdateWorkspaceProjectErrors, UpdateWorkspaceProjectResponses } from './types.gen';
+import type { BlockMachineData, BlockMachineErrors, BlockMachineResponses, CreateKeyData, CreateKeyErrors, CreateKeyResponses, DeleteCatalogModelData, DeleteCatalogModelErrors, DeleteCatalogModelResponses, DeleteFleetPackageData, DeleteFleetPackageResponses, DeleteGpuPoolData, DeleteGpuPoolErrors, DeleteGpuPoolResponses, DeleteKeyData, DeleteKeyErrors, DeleteKeyResponses, DeleteTreeProjectData, DeleteTreeProjectErrors, DeleteTreeProjectResponses, DeleteWorkspaceProjectData, DeleteWorkspaceProjectErrors, DeleteWorkspaceProjectResponses, ForgetMachineData, ForgetMachineErrors, ForgetMachineResponses, GetFleetData, GetFleetErrors, GetFleetInferenceData, GetFleetInferenceErrors, GetFleetInferenceResponses, GetFleetInventoryCsvData, GetFleetInventoryCsvResponses, GetFleetPackageData, GetFleetPackageErrors, GetFleetPackageResponses, GetFleetProfileData, GetFleetProfileResponses, GetFleetReadmeData, GetFleetReadmeResponses, GetFleetResponses, GetGpuFleetData, GetGpuFleetErrors, GetGpuFleetResponses, GetGpuHoursData, GetGpuHoursErrors, GetGpuHoursResponses, GetMeData, GetMeErrors, GetMeResponses, GetModelCatalogData, GetModelCatalogErrors, GetModelCatalogResponses, GetMyWorkspaceUsageData, GetMyWorkspaceUsageErrors, GetMyWorkspaceUsageResponses, GetProjectImageLogData, GetProjectImageLogErrors, GetProjectImageLogResponses, GetTreeCatalogData, GetTreeCatalogErrors, GetTreeCatalogResponses, GetUsageData, GetUsageErrors, GetUsageResponses, GetWorkspaceCatalogData, GetWorkspaceCatalogErrors, GetWorkspaceCatalogResponses, GetWorkspaceStorageData, GetWorkspaceStorageErrors, GetWorkspaceStorageResponses, GetWorkspaceUsageData, GetWorkspaceUsageErrors, GetWorkspaceUsageResponses, ListGpuServersData, ListGpuServersErrors, ListGpuServersResponses, ListKeysData, ListKeysErrors, ListKeysResponses, ListTreeProjectsData, ListTreeProjectsErrors, ListTreeProjectsResponses, ListWorkspaceProjectsData, ListWorkspaceProjectsResponses, PutCatalogModelData, PutCatalogModelErrors, PutCatalogModelResponses, PutGpuNodeData, PutGpuNodeErrors, PutGpuNodeResponses, PutGpuPoolData, PutGpuPoolErrors, PutGpuPoolResponses, PutTierModelData, PutTierModelErrors, PutTierModelResponses, PutTreeProjectData, PutTreeProjectErrors, PutTreeProjectResponses, RebuildProjectImageData, RebuildProjectImageErrors, RebuildProjectImageResponses, SetDefaultProjectData, SetDefaultProjectErrors, SetDefaultProjectResponses, StopGpuServerData, StopGpuServerErrors, StopGpuServerResponses, UnblockMachineData, UnblockMachineErrors, UnblockMachineResponses, UpdateWorkspaceProjectData, UpdateWorkspaceProjectErrors, UpdateWorkspaceProjectResponses } from './types.gen';
 export type Options<TData extends TDataShape = TDataShape, ThrowOnError extends boolean = boolean, TResponse = unknown> = Options2<TData, ThrowOnError, TResponse> & {
     /**
      * You can provide a client instance returned by `createClient()` instead of
@@ -97,6 +97,30 @@ export declare const listGpuServers: <ThrowOnError extends boolean = false>(opti
  * Named servers are removed after stopping; the home stays. The default server has no name and is addressed as "-".
  */
 export declare const stopGpuServer: <ThrowOnError extends boolean = false>(options: Options<StopGpuServerData, ThrowOnError>) => RequestResult<StopGpuServerResponses, StopGpuServerErrors, ThrowOnError>;
+/**
+ * Catalogue of the platform's own models
+ *
+ * For admins: the models the platform runs itself on its GPUs (the inference gap fillers), with their vLLM arguments, resources and needed GPU memory, the default model of each GPU tier, and how their replicas are doing (ready, starting, waiting for a GPU, restarts and the last problem).
+ */
+export declare const getModelCatalog: <ThrowOnError extends boolean = false>(options?: Options<GetModelCatalogData, ThrowOnError>) => RequestResult<GetModelCatalogResponses, GetModelCatalogErrors, ThrowOnError>;
+/**
+ * Delete a model
+ *
+ * For admins: only a model no tier has as default and no node has chosen.
+ */
+export declare const deleteCatalogModel: <ThrowOnError extends boolean = false>(options: Options<DeleteCatalogModelData, ThrowOnError>) => RequestResult<DeleteCatalogModelResponses, DeleteCatalogModelErrors, ThrowOnError>;
+/**
+ * Add or change a model
+ *
+ * For admins: a model of the catalogue. The gap fillers that run it are recreated with it within a minute.
+ */
+export declare const putCatalogModel: <ThrowOnError extends boolean = false>(options: Options<PutCatalogModelData, ThrowOnError>) => RequestResult<PutCatalogModelResponses, PutCatalogModelErrors, ThrowOnError>;
+/**
+ * Default model of a GPU tier
+ *
+ * For admins: the gap filler model of every node of the tier that has not chosen another; empty = none.
+ */
+export declare const putTierModel: <ThrowOnError extends boolean = false>(options: Options<PutTierModelData, ThrowOnError>) => RequestResult<PutTierModelResponses, PutTierModelErrors, ThrowOnError>;
 /**
  * The caller's API keys
  */

package/dist/types.gen.d.ts

@@ -79,6 +79,29 @@ export type FleetProfile = {
     weight?: number;
     wiredLimitMb?: number;
 };
+export type GpuCatalogModel = {
+    /**
+     * vLLM arguments after --model and --port
+     */
+    args?: Array<string>;
+    /**
+     * Resources of the replica's container; empty = 1 CPU, 3–5 GiB, 1 GPU.
+     */
+    cpu?: string;
+    key?: string;
+    memory_limit?: string;
+    memory_request?: string;
+    /**
+     * GPU memory it needs; nodes with less are not offered it
+     */
+    min_gb?: number;
+    /**
+     * Hugging Face repository, also the served name
+     */
+    model?: string;
+    updated_at?: string;
+    updated_by?: string;
+};
 export type GpuClassInfo = {
     /**
      * GPUs of the ready, schedulable assigned nodes
@@ -130,9 +153,18 @@ export type GpuGpuNode = {
 export type GpuInferencePod = {
     gpu_class?: string;
     model?: string;
+    /**
+     * ModelKey is the catalogue model of a gap filler (label dhbw.cloud/inference-model); empty for the base load.
+     */
+    model_key?: string;
     node?: string;
     pod?: string;
+    /**
+     * Problem is why the container is not running or last ended, e.g. "CrashLoopBackOff" or "OOMKilled".
+     */
+    problem?: string;
     replica?: string;
+    restarts?: number;
     role?: 'gap-filler' | 'base-load';
     /**
      * start on the node, or creation while it still waits
@@ -149,6 +181,35 @@ export type GpuInferencePool = {
     pods?: Array<GpuInferencePod>;
     url?: string;
 };
+export type GpuModelCatalog = {
+    enabled?: boolean;
+    models?: Array<GpuCatalogModel>;
+    states?: {
+        [key: string]: GpuModelState;
+    };
+    tier_models?: {
+        [key: string]: string;
+    };
+    tiers?: Array<GpuTier>;
+};
+export type GpuModelChoice = {
+    key?: string;
+    min_gb?: number;
+};
+export type GpuModelState = {
+    /**
+     * the last failure, e.g. CrashLoopBackOff: Error
+     */
+    problem?: string;
+    ready?: number;
+    replicas?: number;
+    restarts?: number;
+    starting?: number;
+    /**
+     * for a GPU (a workspace has it)
+     */
+    waiting?: number;
+};
 export type GpuNodeAssignment = {
     available_until?: string;
     /**
@@ -333,9 +394,9 @@ export type WebserverErrorResponse = {
 export type WebserverGpuFleetResponse = {
     classes?: Array<GpuClassInfo>;
     /**
-     * Models are the gap filler models a node can run (keys), TierModels the default of each tier.
+     * Models are the gap filler models of the catalogue (a node is offered those whose min_gb its GPU memory reaches), TierModels the default of each tier.
      */
-    models?: Array<string>;
+    models?: Array<GpuModelChoice>;
     nodes?: Array<GpuGpuNode>;
     pools?: Array<GpuPool>;
     tier_models?: {
@@ -386,6 +447,9 @@ export type WebserverMeResponse = {
      */
     workspace_projects?: number;
 };
+export type WebserverTierModelRequest = {
+    model?: string;
+};
 export type WebserverTreeCatalog = {
     bases?: Array<WorkspaceBase>;
     gpuClasses?: Array<WorkspaceGpuClass>;
@@ -1053,6 +1117,119 @@ export type StopGpuServerResponses = {
     204: void;
 };
 export type StopGpuServerResponse = StopGpuServerResponses[keyof StopGpuServerResponses];
+export type GetModelCatalogData = {
+    body?: never;
+    path?: never;
+    query?: never;
+    url: '/v1/inference/models';
+};
+export type GetModelCatalogErrors = {
+    /**
+     * Forbidden
+     */
+    403: WebserverErrorResponse;
+};
+export type GetModelCatalogError = GetModelCatalogErrors[keyof GetModelCatalogErrors];
+export type GetModelCatalogResponses = {
+    /**
+     * OK
+     */
+    200: GpuModelCatalog;
+};
+export type GetModelCatalogResponse = GetModelCatalogResponses[keyof GetModelCatalogResponses];
+export type DeleteCatalogModelData = {
+    body?: never;
+    path: {
+        /**
+         * model key
+         */
+        key: string;
+    };
+    query?: never;
+    url: '/v1/inference/models/{key}';
+};
+export type DeleteCatalogModelErrors = {
+    /**
+     * Bad Request
+     */
+    400: WebserverErrorResponse;
+    /**
+     * Forbidden
+     */
+    403: WebserverErrorResponse;
+};
+export type DeleteCatalogModelError = DeleteCatalogModelErrors[keyof DeleteCatalogModelErrors];
+export type DeleteCatalogModelResponses = {
+    /**
+     * OK
+     */
+    200: GpuModelCatalog;
+};
+export type DeleteCatalogModelResponse = DeleteCatalogModelResponses[keyof DeleteCatalogModelResponses];
+export type PutCatalogModelData = {
+    /**
+     * model
+     */
+    body: GpuCatalogModel;
+    path: {
+        /**
+         * model key (lower case, digits, dot, dash)
+         */
+        key: string;
+    };
+    query?: never;
+    url: '/v1/inference/models/{key}';
+};
+export type PutCatalogModelErrors = {
+    /**
+     * Bad Request
+     */
+    400: WebserverErrorResponse;
+    /**
+     * Forbidden
+     */
+    403: WebserverErrorResponse;
+};
+export type PutCatalogModelError = PutCatalogModelErrors[keyof PutCatalogModelErrors];
+export type PutCatalogModelResponses = {
+    /**
+     * OK
+     */
+    200: GpuModelCatalog;
+};
+export type PutCatalogModelResponse = PutCatalogModelResponses[keyof PutCatalogModelResponses];
+export type PutTierModelData = {
+    /**
+     * model
+     */
+    body: WebserverTierModelRequest;
+    path: {
+        /**
+         * tier (klein, mittel, gross)
+         */
+        tier: string;
+    };
+    query?: never;
+    url: '/v1/inference/tiers/{tier}';
+};
+export type PutTierModelErrors = {
+    /**
+     * Bad Request
+     */
+    400: WebserverErrorResponse;
+    /**
+     * Forbidden
+     */
+    403: WebserverErrorResponse;
+};
+export type PutTierModelError = PutTierModelErrors[keyof PutTierModelErrors];
+export type PutTierModelResponses = {
+    /**
+     * OK
+     */
+    200: GpuModelCatalog;
+};
+export type PutTierModelResponse = PutTierModelResponses[keyof PutTierModelResponses];
 export type ListKeysData = {
     body?: never;
     path?: never;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@dhbw-cloud/llm-client",
-  "version": "0.2.16",
+  "version": "0.2.17",
   "description": "Generated TypeScript client for the DHBW Cloud llm-management-api.",
   "license": "Apache-2.0",
   "type": "module",