@dhbw-cloud/llm-client 0.2.11 → 0.2.16

package/dist/index.d.ts

@@ -1,2 +1,2 @@
-export { blockMachine, createKey, deleteFleetPackage, deleteKey, deleteTreeProject, deleteWorkspaceProject, forgetMachine, getFleet, getFleetInference, getFleetInventoryCsv, getFleetPackage, getFleetProfile, getFleetReadme, getGpuHours, getMe, getMyWorkspaceUsage, getProjectImageLog, getTreeCatalog, getUsage, getWorkspaceCatalog, getWorkspaceStorage, getWorkspaceUsage, listGpuServers, listKeys, listTreeProjects, listWorkspaceProjects, type Options, 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, 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, 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, GpuInferencePod, GpuInferencePool, GpuServer, KeysKeyView, KeysProjectUsage, KeysUsage, ListGpuServersData, ListGpuServersError, ListGpuServersErrors, ListGpuServersResponse, ListGpuServersResponses, ListKeysData, ListKeysError, ListKeysErrors, ListKeysResponse, ListKeysResponses, ListTreeProjectsData, ListTreeProjectsError, ListTreeProjectsErrors, ListTreeProjectsResponse, ListTreeProjectsResponses, ListWorkspaceProjectsData, ListWorkspaceProjectsResponse, ListWorkspaceProjectsResponses, 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, WebserverMeResponse, WebserverTreeCatalog, WorkspaceBudget, WorkspaceBudgetStatus, WorkspaceConfig, WorkspaceImageBuild, WorkspaceLimits, WorkspaceMember, WorkspaceMine, WorkspaceProfile, WorkspaceProject, WorkspaceStorageDir, WorkspaceStorageReport, WorkspaceTreeBudget, WorkspaceTreeProject, WorkspaceTreeStatus, WorkspaceTreeUse, WorkspaceValkeyMemory } from './types.gen';
+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';

package/dist/index.mjs

@@ -837,6 +837,34 @@ var unblockMachine = (options) => (options.client ?? client).post({
   url: "/v1/fleet/{serial}/unblock",
   ...options
 });
+var getGpuFleet = (options) => (options?.client ?? client).get({
+  security: [{ name: "Authorization", type: "apiKey" }],
+  url: "/v1/gpu/fleet",
+  ...options
+});
+var putGpuNode = (options) => (options.client ?? client).put({
+  security: [{ name: "Authorization", type: "apiKey" }],
+  url: "/v1/gpu/fleet/nodes/{name}",
+  ...options,
+  headers: {
+    "Content-Type": "application/json",
+    ...options.headers
+  }
+});
+var deleteGpuPool = (options) => (options.client ?? client).delete({
+  security: [{ name: "Authorization", type: "apiKey" }],
+  url: "/v1/gpu/fleet/pools/{id}",
+  ...options
+});
+var putGpuPool = (options) => (options.client ?? client).put({
+  security: [{ name: "Authorization", type: "apiKey" }],
+  url: "/v1/gpu/fleet/pools/{id}",
+  ...options,
+  headers: {
+    "Content-Type": "application/json",
+    ...options.headers
+  }
+});
 var listGpuServers = (options) => (options?.client ?? client).get({
   security: [{ name: "Authorization", type: "apiKey" }],
   url: "/v1/gpu/servers",
@@ -968,6 +996,7 @@ export {
   client,
   createKey,
   deleteFleetPackage,
+  deleteGpuPool,
   deleteKey,
   deleteTreeProject,
   deleteWorkspaceProject,
@@ -978,6 +1007,7 @@ export {
   getFleetPackage,
   getFleetProfile,
   getFleetReadme,
+  getGpuFleet,
   getGpuHours,
   getMe,
   getMyWorkspaceUsage,
@@ -991,6 +1021,8 @@ export {
   listKeys,
   listTreeProjects,
   listWorkspaceProjects,
+  putGpuNode,
+  putGpuPool,
   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, 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, 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, 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, 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';
 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
@@ -63,6 +63,30 @@ export declare const blockMachine: <ThrowOnError extends boolean = false>(option
  * Unblock a machine
  */
 export declare const unblockMachine: <ThrowOnError extends boolean = false>(options: Options<UnblockMachineData, ThrowOnError>) => RequestResult<UnblockMachineResponses, UnblockMachineErrors, ThrowOnError>;
+/**
+ * GPU administration
+ *
+ * For admins: the GPU nodes of the cluster (type, GPU memory, ready) with their pool and tier, the pools, the tiers and the GPU classes. A node without pool runs no workspaces and no jobs. The tree grants classes (ws-gpus-<class>, ws-gpu-hours-<class>), so new nodes of an existing class need nothing there.
+ */
+export declare const getGpuFleet: <ThrowOnError extends boolean = false>(options?: Options<GetGpuFleetData, ThrowOnError>) => RequestResult<GetGpuFleetResponses, GetGpuFleetErrors, ThrowOnError>;
+/**
+ * Assign a GPU node
+ *
+ * For admins: puts a GPU node into a pool and tier (the class is created on first use) or, with an empty pool, takes it out. The cluster follows within a minute: node labels and the Kueue queues of the classes. Running pods stay where they are.
+ */
+export declare const putGpuNode: <ThrowOnError extends boolean = false>(options: Options<PutGpuNodeData, ThrowOnError>) => RequestResult<PutGpuNodeResponses, PutGpuNodeErrors, ThrowOnError>;
+/**
+ * Delete a GPU pool
+ *
+ * For admins: only a pool without classes (never used by a node) can go.
+ */
+export declare const deleteGpuPool: <ThrowOnError extends boolean = false>(options: Options<DeleteGpuPoolData, ThrowOnError>) => RequestResult<DeleteGpuPoolResponses, DeleteGpuPoolErrors, ThrowOnError>;
+/**
+ * Create or rename a GPU pool
+ *
+ * For admins: a pool is who GPUs are for — "allgemein", or GPUs of one group, granted in the tree only to that group's budgets. The id is fixed once created (it becomes part of the tree resources).
+ */
+export declare const putGpuPool: <ThrowOnError extends boolean = false>(options: Options<PutGpuPoolData, ThrowOnError>) => RequestResult<PutGpuPoolResponses, PutGpuPoolErrors, ThrowOnError>;
 /**
  * My JupyterHub servers
  */

package/dist/types.gen.d.ts

@@ -79,6 +79,54 @@ export type FleetProfile = {
     weight?: number;
     wiredLimitMb?: number;
 };
+export type GpuClassInfo = {
+    /**
+     * GPUs of the ready, schedulable assigned nodes
+     */
+    gpus?: number;
+    id?: string;
+    name?: string;
+    /**
+     * assigned nodes, ready or not
+     */
+    nodes?: number;
+    pool?: string;
+    pool_name?: string;
+    /**
+     * GPU types of its nodes, e.g. "NVIDIA-L40S-12Q"
+     */
+    products?: string;
+    /**
+     * the id from before pools this class took over
+     */
+    replaces?: string;
+    tier?: string;
+    tier_name?: string;
+};
+export type GpuGpuNode = {
+    /**
+     * Assignment; nil = not assigned: it runs no workspaces and no jobs.
+     */
+    assignment?: GpuNodeAssignment;
+    class?: string;
+    cpu?: number;
+    gpus?: number;
+    /**
+     * GPU memory per GPU
+     */
+    memory_gb?: number;
+    memory_gi?: number;
+    name?: string;
+    product?: string;
+    /**
+     * ready and schedulable
+     */
+    ready?: boolean;
+    /**
+     * SuggestedTier is the tier its GPU memory suggests.
+     */
+    suggested_tier?: string;
+};
 export type GpuInferencePod = {
     gpu_class?: string;
     model?: string;
@@ -101,6 +149,27 @@ export type GpuInferencePool = {
     pods?: Array<GpuInferencePod>;
     url?: string;
 };
+export type GpuNodeAssignment = {
+    available_until?: string;
+    /**
+     * GapModel: the inference gap filler on this node, a key of the configured models; "" = the tier's default, "-" = none.
+     */
+    gap_model?: string;
+    node?: string;
+    pool?: string;
+    tier?: string;
+    updated_at?: string;
+    updated_by?: string;
+};
+export type GpuPool = {
+    created_at?: string;
+    id?: string;
+    /**
+     * idle GPUs of this pool may be used by the other pools of the tier (not yet in effect)
+     */
+    lend?: boolean;
+    name?: string;
+};
 export type GpuServer = {
     image?: string;
     name?: string;
@@ -109,6 +178,14 @@ export type GpuServer = {
     ready?: boolean;
     url?: string;
 };
+export type GpuTier = {
+    id?: string;
+    /**
+     * a node is suggested this tier from this much GPU memory on
+     */
+    min_gb?: number;
+    name?: string;
+};
 export type KeysKeyView = {
     budget?: number;
     budget_duration?: string;
@@ -203,6 +280,10 @@ export type UsageTotal = {
     days?: number;
     gpu_class?: string;
     gpu_hours?: number;
+    /**
+     * GPUProduct is the GPU type used (nvidia.com/gpu.product of the node).
+     */
+    gpu_product?: string;
     /**
      * sum over days of the day's home size
      */
@@ -249,6 +330,32 @@ export type WebserverErrorResponse = {
      */
     error?: string;
 };
+export type WebserverGpuFleetResponse = {
+    classes?: Array<GpuClassInfo>;
+    /**
+     * Models are the gap filler models a node can run (keys), TierModels the default of each tier.
+     */
+    models?: Array<string>;
+    nodes?: Array<GpuGpuNode>;
+    pools?: Array<GpuPool>;
+    tier_models?: {
+        [key: string]: string;
+    };
+    tiers?: Array<GpuTier>;
+};
+export type WebserverGpuNodeRequest = {
+    available_until?: string;
+    /**
+     * GapModel: inference gap filler; "" = the tier's default, "-" = none.
+     */
+    gap_model?: string;
+    pool?: string;
+    tier?: string;
+};
+export type WebserverGpuPoolRequest = {
+    lend?: boolean;
+    name?: string;
+};
 export type WebserverMeResponse = {
     /**
      * ActingAs is the person this request acted for (the e-mail above); empty without X-Act-As.
@@ -280,12 +387,19 @@ export type WebserverMeResponse = {
     workspace_projects?: number;
 };
 export type WebserverTreeCatalog = {
-    gpuClasses?: Array<string>;
+    bases?: Array<WorkspaceBase>;
+    gpuClasses?: Array<WorkspaceGpuClass>;
     llm_base_models?: Array<string>;
     llm_default_rpm?: number;
     llm_default_tpm?: number;
     profiles?: Array<WorkspaceProfile>;
 };
+export type WorkspaceBase = {
+    default?: boolean;
+    id?: string;
+    image?: string;
+    name?: string;
+};
 export type WorkspaceBudget = {
     amount?: number;
     /**
@@ -311,9 +425,28 @@ export type WorkspaceBudgetStatus = {
     window?: '24h' | '7d' | '30d' | 'total';
 };
 export type WorkspaceConfig = {
-    gpuClasses?: Array<string>;
+    bases?: Array<WorkspaceBase>;
+    gpuClasses?: Array<WorkspaceGpuClass>;
     profiles?: Array<WorkspaceProfile>;
 };
+export type WorkspaceGpuClass = {
+    /**
+     * GPUs of its ready nodes right now
+     */
+    gpus?: number;
+    id?: string;
+    name?: string;
+    nodes?: number;
+    pool?: string;
+    pool_name?: string;
+    products?: string;
+    /**
+     * Replaces is the id from before pools this class took over; the tree renames its resources once.
+     */
+    replaces?: string;
+    tier?: string;
+    tier_name?: string;
+};
 export type WorkspaceImageBuild = {
     branch?: string;
     commit?: string;
@@ -338,7 +471,7 @@ export type WorkspaceLimits = {
     servers_per_person?: number;
 };
 export type WorkspaceMember = {
-    role?: 'member' | 'manager';
+    role?: 'reader' | 'member' | 'manager';
     token?: string;
 };
 export type WorkspaceMine = {
@@ -355,10 +488,6 @@ export type WorkspaceMine = {
      * after this day nobody can start in the project
      */
     ends_at?: string;
-    /**
-     * W3: any image, with or without GPU (profiles free, free-gpu)
-     */
-    free_images?: boolean;
     gpu_classes?: Array<string>;
     /**
      * GPUsByClass is how many GPUs of a class the whole project may use at the same time; a class without an entry has no such limit.
@@ -367,13 +496,21 @@ export type WorkspaceMine = {
         [key: string]: number;
     };
     id?: number;
+    /**
+     * The project's environment, run by every profile but the free images (docs/bauplan.md 10.12; Service.EffectiveImage): ImageBase is the chosen base (Config.Bases, empty = the default). ProjectImage is a finished image derived from a base (label BaseLabel), or, with ImageGitURL, the last successful build from that repository on ImageBase; empty = the base itself.
+     */
+    image_base?: string;
     /**
      * ImageBuild is the state of the project image's build from Git (ImageGitURL); nil = none.
      */
     image_build?: WorkspaceImageBuild;
+    /**
+     * ImageDerived tells whether an existing ProjectImage is derived from a base (label BaseLabel, looked up on saving). Any image may be the project's; one that is not derived runs as it is (its own jupyterhub-singleuser), without the GPU kernel (Derived).
+     */
+    image_derived?: boolean;
     image_git_ref?: string;
     /**
-     * ImageGitURL builds ProjectImage from a public repository (requirements.txt, environment.yml, apt.txt, postBuild) on notebook-base, rebuilt on its new releases; ImageGitRef is a branch, tag or commit (empty = default branch).
+     * ImageGitURL builds ProjectImage from a public repository (requirements.txt, environment.yml, apt.txt, postBuild) on ImageBase, rebuilt on its new releases; ImageGitRef is a branch, tag or commit (empty = default branch).
      */
     image_git_url?: string;
     limits?: WorkspaceLimits;
@@ -393,11 +530,8 @@ export type WorkspaceMine = {
     members?: Array<WorkspaceMember>;
     name?: string;
     profiles?: Array<string>;
-    /**
-     * ProjectImage is the image of the profiles "CPU + GPU-Kernel" and "Projekt-Image (GPU)" instead of notebook-base (docs/bauplan.md 10.12): a finished image derived from notebook-base (label BaseLabel), or, with ImageGitURL, the last successful build from that repository; empty = notebook-base.
-     */
     project_image?: string;
-    role?: 'member' | 'manager' | 'admin';
+    role?: 'reader' | 'member' | 'manager' | 'admin';
     /**
      * short name, also the pod label dhbw.cloud/project
      */
@@ -414,7 +548,7 @@ export type WorkspaceMine = {
     updated_by?: string;
 };
 export type WorkspaceProfile = {
-    gpu_class?: string;
+    gpu?: boolean;
     name?: string;
     slug?: string;
 };
@@ -431,10 +565,6 @@ export type WorkspaceProject = {
      * after this day nobody can start in the project
      */
     ends_at?: string;
-    /**
-     * W3: any image, with or without GPU (profiles free, free-gpu)
-     */
-    free_images?: boolean;
     gpu_classes?: Array<string>;
     /**
      * GPUsByClass is how many GPUs of a class the whole project may use at the same time; a class without an entry has no such limit.
@@ -443,9 +573,17 @@ export type WorkspaceProject = {
         [key: string]: number;
     };
     id?: number;
+    /**
+     * The project's environment, run by every profile but the free images (docs/bauplan.md 10.12; Service.EffectiveImage): ImageBase is the chosen base (Config.Bases, empty = the default). ProjectImage is a finished image derived from a base (label BaseLabel), or, with ImageGitURL, the last successful build from that repository on ImageBase; empty = the base itself.
+     */
+    image_base?: string;
+    /**
+     * ImageDerived tells whether an existing ProjectImage is derived from a base (label BaseLabel, looked up on saving). Any image may be the project's; one that is not derived runs as it is (its own jupyterhub-singleuser), without the GPU kernel (Derived).
+     */
+    image_derived?: boolean;
     image_git_ref?: string;
     /**
-     * ImageGitURL builds ProjectImage from a public repository (requirements.txt, environment.yml, apt.txt, postBuild) on notebook-base, rebuilt on its new releases; ImageGitRef is a branch, tag or commit (empty = default branch).
+     * ImageGitURL builds ProjectImage from a public repository (requirements.txt, environment.yml, apt.txt, postBuild) on ImageBase, rebuilt on its new releases; ImageGitRef is a branch, tag or commit (empty = default branch).
      */
     image_git_url?: string;
     limits?: WorkspaceLimits;
@@ -465,9 +603,6 @@ export type WorkspaceProject = {
     members?: Array<WorkspaceMember>;
     name?: string;
     profiles?: Array<string>;
-    /**
-     * ProjectImage is the image of the profiles "CPU + GPU-Kernel" and "Projekt-Image (GPU)" instead of notebook-base (docs/bauplan.md 10.12): a finished image derived from notebook-base (label BaseLabel), or, with ImageGitURL, the last successful build from that repository; empty = notebook-base.
-     */
     project_image?: string;
     /**
      * short name, also the pod label dhbw.cloud/project
@@ -760,6 +895,119 @@ export type UnblockMachineResponses = {
     204: void;
 };
 export type UnblockMachineResponse = UnblockMachineResponses[keyof UnblockMachineResponses];
+export type GetGpuFleetData = {
+    body?: never;
+    path?: never;
+    query?: never;
+    url: '/v1/gpu/fleet';
+};
+export type GetGpuFleetErrors = {
+    /**
+     * Forbidden
+     */
+    403: WebserverErrorResponse;
+};
+export type GetGpuFleetError = GetGpuFleetErrors[keyof GetGpuFleetErrors];
+export type GetGpuFleetResponses = {
+    /**
+     * OK
+     */
+    200: WebserverGpuFleetResponse;
+};
+export type GetGpuFleetResponse = GetGpuFleetResponses[keyof GetGpuFleetResponses];
+export type PutGpuNodeData = {
+    /**
+     * assignment
+     */
+    body: WebserverGpuNodeRequest;
+    path: {
+        /**
+         * node name
+         */
+        name: string;
+    };
+    query?: never;
+    url: '/v1/gpu/fleet/nodes/{name}';
+};
+export type PutGpuNodeErrors = {
+    /**
+     * Bad Request
+     */
+    400: WebserverErrorResponse;
+    /**
+     * Forbidden
+     */
+    403: WebserverErrorResponse;
+};
+export type PutGpuNodeError = PutGpuNodeErrors[keyof PutGpuNodeErrors];
+export type PutGpuNodeResponses = {
+    /**
+     * OK
+     */
+    200: WebserverGpuFleetResponse;
+};
+export type PutGpuNodeResponse = PutGpuNodeResponses[keyof PutGpuNodeResponses];
+export type DeleteGpuPoolData = {
+    body?: never;
+    path: {
+        /**
+         * pool id
+         */
+        id: string;
+    };
+    query?: never;
+    url: '/v1/gpu/fleet/pools/{id}';
+};
+export type DeleteGpuPoolErrors = {
+    /**
+     * Bad Request
+     */
+    400: WebserverErrorResponse;
+    /**
+     * Forbidden
+     */
+    403: WebserverErrorResponse;
+};
+export type DeleteGpuPoolError = DeleteGpuPoolErrors[keyof DeleteGpuPoolErrors];
+export type DeleteGpuPoolResponses = {
+    /**
+     * OK
+     */
+    200: WebserverGpuFleetResponse;
+};
+export type DeleteGpuPoolResponse = DeleteGpuPoolResponses[keyof DeleteGpuPoolResponses];
+export type PutGpuPoolData = {
+    /**
+     * pool
+     */
+    body: WebserverGpuPoolRequest;
+    path: {
+        /**
+         * pool id (2–20 lower-case letters or digits)
+         */
+        id: string;
+    };
+    query?: never;
+    url: '/v1/gpu/fleet/pools/{id}';
+};
+export type PutGpuPoolErrors = {
+    /**
+     * Bad Request
+     */
+    400: WebserverErrorResponse;
+    /**
+     * Forbidden
+     */
+    403: WebserverErrorResponse;
+};
+export type PutGpuPoolError = PutGpuPoolErrors[keyof PutGpuPoolErrors];
+export type PutGpuPoolResponses = {
+    /**
+     * OK
+     */
+    200: WebserverGpuFleetResponse;
+};
+export type PutGpuPoolResponse = PutGpuPoolResponses[keyof PutGpuPoolResponses];
 export type ListGpuServersData = {
     body?: never;
     path?: never;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@dhbw-cloud/llm-client",
-  "version": "0.2.11",
+  "version": "0.2.16",
   "description": "Generated TypeScript client for the DHBW Cloud llm-management-api.",
   "license": "Apache-2.0",
   "type": "module",