@qvac/sdk 0.7.1 → 0.8.1

package/NOTICE

@@ -37,12 +37,36 @@ Additional requirements per license Section 1.b.i:
   must prominently display "Built with Llama" on a related website,
   user interface, blog post, about page, or product documentation.
 
+# Model Use Restrictions (Open RAIL++-M)
+
+As a user of the QVAC SDK and its associated model registry, you agree not to use the Model or its Derivatives:
+
+1. **Illegal Acts:** In any way that violates any applicable national, federal, state, local or international law or regulation.
+2. **Harm to Minors:** For the purpose of exploiting, harming or attempting to exploit or harm minors in any way.
+3. **Malicious Content:** To generate or disseminate verifiably false information and/or content with the purpose of harming others.
+4. **Personal Data:** To generate or disseminate personal identifiable information that can be used to harm an individual.
+5. **Defamation & Harassment:** To defame, disparage or otherwise harass others.
+6. **Automated Decisions:** For fully automated decision making that adversely impacts an individual's legal rights or otherwise creates or modifies a binding, enforceable obligation.
+7. **Discrimination:** For any use intended to or which has the effect of discriminating against or harming individuals or groups based on online or offline social behavior or known or predicted personal or personality characteristics.
+8. **Exploitation:** To exploit any of the vulnerabilities of a specific group of persons based on their age, social, physical or mental characteristics, in order to materially distort the behavior of a person pertaining to such group in a manner that causes or is likely to cause that person or another person physical or psychological harm.
+9. **Medical Advice:** To provide medical advice and medical results interpretation.
+10. **Justice & Law Enforcement:** To generate or disseminate information for use by administration of justice, law enforcement, immigration or asylum processes, such as predicting an individual will commit fraud/crime commitment (e.g. by text profiling, drawing causal relationships between assertions made in documents, indiscriminate and untargeted scraping).
+
+---
+*Note: These restrictions must be included in any distribution of the Model or Derivatives thereof to your end-users.*
+
 =========================================================================
 Third-Party Model Licenses
 =========================================================================
 
 --- apache-2.0 (Apache License 2.0) ---
 
+  crnn_mobilenet_v3_small
+    https://github.com/felixdittrich92/OnnxTR
+  db_mobilenet_v3_large
+    https://github.com/felixdittrich92/OnnxTR
+  db_resnet50
+    https://github.com/felixdittrich92/OnnxTR
   de-tiny-ggml-model-f16
     https://huggingface.co/primeline/whisper-tiny-german-1224
   detector_craft
@@ -51,6 +75,10 @@ Third-Party Model Licenses
     https://huggingface.co/jmb95/laser-dolphin-mixtral-2x7b-dpo-GGUF
   en-base-ggml-model-f16
     https://huggingface.co/openai/whisper-base
+  FLUX.2-klein-4B
+    https://huggingface.co/black-forest-labs/FLUX.2-klein-4B
+  FLUX.2-klein-4B-GGUF
+    https://huggingface.co/unsloth/FLUX.2-klein-4B-GGUF
   fr-base-ggml-model-f16
     https://huggingface.co/personalizedrefrigerator/whisper-base-fr
   fr-tiny-ggml-model-f16
@@ -95,6 +123,8 @@ Third-Party Model Licenses
     https://huggingface.co/Helsinki-NLP/opus-mt-roa-en
   ggml-opus-roa-en
     https://huggingface.co/Helsinki-NLP/opus-mt-roa-en
+  gpt-oss-120b-Q4_K_M-00001-of-00002
+    https://huggingface.co/unsloth/gpt-oss-120b-GGUF
   gpt-oss-20b-GGUF
     https://huggingface.co/unsloth/gpt-oss-20b-GGUF
   it-base-ggml-model-f16
@@ -107,10 +137,14 @@ Third-Party Model Licenses
     https://huggingface.co/CheeLi03/whisper-tiny-ja-puct-4k
   laser-dolphin-mixtral-2x7b-dpo-GGUF
     https://huggingface.co/jmb95/laser-dolphin-mixtral-2x7b-dpo-GGUF
+  LightOnOCR-2-1B-ocr-soup-GGUF
+    https://huggingface.co/noctrex/LightOnOCR-2-1B-ocr-soup-GGUF
   Llama_3.2_1B_Intruct_Tool_Calling_V2-GGUF
     https://huggingface.co/mav23/Llama_3.2_1B_Intruct_Tool_Calling_V2-GGUF
   nb-tiny-ggml-model
     https://huggingface.co/NbAiLab/nb-whisper-tiny
+  parseq
+    https://github.com/felixdittrich92/OnnxTR
   pt-base-ggml-model-f16
     https://huggingface.co/mariana-coelho-9/whisper-base-pt
   Qwen3-0.6B-GGUF
@@ -119,6 +153,8 @@ Third-Party Model Licenses
     https://huggingface.co/unsloth/Qwen3-1.7B-GGUF
   Qwen3-1.7B-Q4_0-00001-of-00005
     https://huggingface.co/unsloth/Qwen3-1.7B-GGUF
+  Qwen3-4B-GGUF
+    https://huggingface.co/unsloth/Qwen3-4B-GGUF
   Qwen3-4B-Q4_0-00001-of-00005
     https://huggingface.co/unsloth/Qwen3-4B-GGUF
   Qwen3-4B-Q4_K_M
@@ -168,8 +204,16 @@ Third-Party Model Licenses
 
 --- cc-by-4.0 (Creative Commons Attribution 4.0 International) ---
 
+  AfriqueGemma-4B-GGUF
+    https://huggingface.co/mradermacher/AfriqueGemma-4B-GGUF
+  decoder_joint_int8
+    https://huggingface.co/nasedkinpv/parakeet-tdt-0.6b-v3-onnx-int8
   decoder_joint-model
     https://huggingface.co/istupakov/parakeet-tdt-0.6b-v3-onnx
+  diar_streaming_sortformer_4spk-v2-onnx
+    https://huggingface.co/cgus/diar_streaming_sortformer_4spk-v2-onnx
+  encoder_int8
+    https://huggingface.co/nasedkinpv/parakeet-tdt-0.6b-v3-onnx-int8
   encoder-model
     https://huggingface.co/istupakov/parakeet-tdt-0.6b-v3-onnx
   ggml-opus-en-de
@@ -180,10 +224,16 @@ Third-Party Model Licenses
     https://huggingface.co/Helsinki-NLP/opus-mt-ru-en
   ggml-opus-zh-en
     https://huggingface.co/Helsinki-NLP/opus-mt-zh-en
+  parakeet-ctc-0.6b-ONNX
+    https://huggingface.co/onnx-community/parakeet-ctc-0.6b-ONNX
+  parakeet-rs
+    https://huggingface.co/altunenes/parakeet-rs
   parakeet-tdt-0.6b-v3-onnx
     https://huggingface.co/istupakov/parakeet-tdt-0.6b-v3-onnx
   preprocessor
     https://huggingface.co/istupakov/parakeet-tdt-0.6b-v3-onnx
+  sortformer_int8
+    https://huggingface.co/nvidia/diar_streaming_sortformer_4spk-v2
 
 --- gemma (Gemma Terms of Use) ---
 
@@ -208,6 +258,12 @@ Third-Party Model Licenses
 
 --- mit (MIT License) ---
 
+  bitnet_b1_58-3B-TQ2_0
+    https://huggingface.co/1bitLLM/bitnet_b1_58-3B
+  bitnet_b1_58-large-TQ2_0
+    https://huggingface.co/1bitLLM/bitnet_b1_58-large
+  bitnet_b1_58-xl-TQ2_0
+    https://huggingface.co/1bitLLM/bitnet_b1_58-xl
   chatterbox-multilingual-ONNX
     https://huggingface.co/onnx-community/chatterbox-multilingual-ONNX
   chatterbox-turbo-ONNX
@@ -287,6 +343,13 @@ Third-Party Model Licenses
   Supertonic-TTS-ONNX
     https://huggingface.co/onnx-community/Supertonic-TTS-ONNX
 
+--- openrail++ ---
+
+  stable-diffusion-v2-1-GGUF
+    https://huggingface.co/gpustack/stable-diffusion-v2-1-GGUF
+  stable-diffusion-xl-base-1.0-GGUF
+    https://huggingface.co/gpustack/stable-diffusion-xl-base-1.0-GGUF
+
 =========================================================================
 Modification Notice
 =========================================================================
@@ -312,26 +375,28 @@ JavaScript Dependencies
   @qvac/dl-base@0.2.0
   @qvac/dl-filesystem@0.2.0
   @qvac/dl-hyperdrive@0.2.0
-  @qvac/embed-llamacpp@0.10.7
+  @qvac/embed-llamacpp@0.12.0
     https://github.com/tetherto/qvac-lib-infer-llamacpp-embed
   @qvac/error@0.1.1
   @qvac/infer-base@0.1.1
   @qvac/infer-base@0.2.2
   @qvac/langdetect-text@0.1.1
-  @qvac/llm-llamacpp@0.8.9
+  @qvac/llm-llamacpp@0.12.2
     https://github.com/tetherto/qvac-lib-infer-llamacpp-llm
   @qvac/logging@0.1.0
-  @qvac/ocr-onnx@0.1.7
+  @qvac/ocr-onnx@0.2.0
     https://github.com/tetherto/qvac-lib-inference-addon-onnx-ocr-fasttext
   @qvac/rag@0.4.2
-  @qvac/registry-client@0.2.0
+  @qvac/registry-client@0.2.1
   @qvac/registry-schema@0.1.1
   @qvac/response@0.1.2
-  @qvac/transcription-whispercpp@0.3.18
-    https://github.com/tetherto/qvac-lib-infer-whispercpp
-  @qvac/translation-nmtcpp@0.3.9
+  @qvac/transcription-parakeet@0.2.0
+    https://github.com/tetherto/qvac
+  @qvac/transcription-whispercpp@0.5.0
+    https://github.com/tetherto/qvac
+  @qvac/translation-nmtcpp@0.6.1
     https://github.com/tetherto/qvac-lib-infer-nmtcpp
-  @qvac/tts-onnx@0.5.4
+  @qvac/tts-onnx@0.6.1
   adaptive-timeout@1.0.1
     https://github.com/holepunchto/adaptive-timeout
   b4a@1.7.5
@@ -502,8 +567,6 @@ JavaScript Dependencies
 
 --- isc (ISC License) ---
 
-  @qvac/util-transcription@0.1.5
-    https://github.com/tetherto/qvac-util-transcription
   bits-to-bytes@1.3.0
     https://github.com/holepunchto/bits-to-bytes
   compact-encoding-bitfield@1.0.0

package/README.md

@@ -167,7 +167,7 @@ bun run examples/path/to/example.ts
 ### Completion
 
 - `llama.cpp` with local files: [`examples/llamacpp-filesystem.ts`](examples/llamacpp-filesystem.ts)
-- `llama.cpp` with Hyperdrive: [`examples/llamacpp-hyperdrive.ts`](examples/llamacpp-hyperdrive.ts)
+- `llama.cpp` with P2P registry: [`examples/llamacpp-p2p.ts`](examples/llamacpp-p2p.ts)
 - `llama.cpp` with HTTP: [`examples/llamacpp-http.ts`](examples/llamacpp-http.ts)
 - `llama.cpp` with tools/function calls: [`examples/llamacpp-native-tools.ts`](examples/llamacpp-native-tools.ts)
 - `llama.cpp` with multimodal inference: [`examples/llamacpp-multimodal.ts`](examples/llamacpp-multimodal.ts)
@@ -180,7 +180,7 @@ bun run examples/path/to/example.ts
 
 ### Embeddings
 
-- Single and batch embeddings: [`examples/embed-hyperdrive.ts`](examples/embed-hyperdrive.ts)
+- Single and batch embeddings: [`examples/embed-p2p.ts`](examples/embed-p2p.ts)
 
 **RAG with HyperDB** (cross-platform):
 
@@ -478,10 +478,10 @@ bun run build  # or `watch` for hotreload
 bun run build:pack
 ```
 
-This outputs a tarball under `dist/qvac-sdk-{version}.tgz` that you can install in your project, e.g.:
+This outputs a tarball under `dist/sdk-{version}.tgz` that you can install in your project, e.g.:
 
 ```bash
-npm i path/to/qvac-sdk-0.3.0.tgz
+npm i path/to/sdk-0.3.0.tgz
 ```
 
 ## More Resources

package/bare-imports.json

@@ -31,5 +31,14 @@
   "readline": {
     "bare": "bare-readline",
     "default": "readline"
+  },
+  "url": {
+    "bare": "bare-url",
+    "default": "url"
+  },
+  "node:url": {
+    "bare": "bare-url",
+    "default": "node:url"
   }
 }
+

package/dist/client/api/cancel.d.ts

@@ -7,6 +7,7 @@ import { type CancelParams } from "../../schemas/index";
  * @param params.modelId - The model ID (required for inference cancellation)
  * @param params.downloadKey - The download key (required for download cancellation)
  * @param params.clearCache - If true, deletes the partial download file (default: false)
+ * @param params.delegate - Delegation target for remote download cancellation (optional)
  * @param params.workspace - The RAG workspace to cancel (optional, defaults to "default")
  * @throws {QvacErrorBase} When the response type is invalid or when the cancellation fails
  *
@@ -23,6 +24,14 @@ import { type CancelParams } from "../../schemas/index";
  * await cancel({ operation: "downloadAsset", downloadKey: "download-key", clearCache: true });
  *
  * @example
+ * // Cancel delegated remote download
+ * await cancel({
+ *   operation: "downloadAsset",
+ *   downloadKey: "download-key",
+ *   delegate: { topic: "topicHex", providerPublicKey: "peerHex" },
+ * });
+ *
+ * @example
  * // Cancel RAG operation on default workspace
  * await cancel({ operation: "rag" });
  *

package/dist/client/api/cancel.js

@@ -9,6 +9,7 @@ import { InvalidResponseError, CancelFailedError } from "../../utils/errors-clie
  * @param params.modelId - The model ID (required for inference cancellation)
  * @param params.downloadKey - The download key (required for download cancellation)
  * @param params.clearCache - If true, deletes the partial download file (default: false)
+ * @param params.delegate - Delegation target for remote download cancellation (optional)
  * @param params.workspace - The RAG workspace to cancel (optional, defaults to "default")
  * @throws {QvacErrorBase} When the response type is invalid or when the cancellation fails
  *
@@ -25,6 +26,14 @@ import { InvalidResponseError, CancelFailedError } from "../../utils/errors-clie
  * await cancel({ operation: "downloadAsset", downloadKey: "download-key", clearCache: true });
  *
  * @example
+ * // Cancel delegated remote download
+ * await cancel({
+ *   operation: "downloadAsset",
+ *   downloadKey: "download-key",
+ *   delegate: { topic: "topicHex", providerPublicKey: "peerHex" },
+ * });
+ *
+ * @example
  * // Cancel RAG operation on default workspace
  * await cancel({ operation: "rag" });
  *

package/dist/client/api/completion-stream.d.ts

@@ -1,8 +1,9 @@
-import { type CompletionClientParams, type CompletionStats, type McpClientInput, type Tool, type ToolCallEvent, type ToolCallWithCall } from "../../schemas/index";
+import { type CompletionClientParams, type CompletionStats, type McpClientInput, type Tool, type ToolCallEvent, type ToolCallWithCall, type RPCOptions } from "../../schemas/index";
 import { type ToolInput } from "../../utils/tool-helpers";
 type CompletionParams = Omit<CompletionClientParams, "tools"> & {
     tools?: Tool[] | ToolInput[];
     mcp?: McpClientInput[];
+    rpcOptions?: RPCOptions;
 };
 /**
  * Generates completion from a language model based on conversation history.

package/dist/client/api/completion-stream.js

@@ -112,8 +112,9 @@ export function completion(params) {
                 kvCache: params.kvCache,
                 tools: allTools.length > 0 ? allTools : undefined,
                 stream: params.stream ?? true,
+                generationParams: params.generationParams,
             };
-            const responses = streamRpc(request);
+            const responses = streamRpc(request, params.rpcOptions);
             for await (const response of responses) {
                 if (response &&
                     typeof response === "object" &&

package/dist/client/api/download-asset.d.ts

@@ -1,4 +1,4 @@
-import { type DownloadAssetOptions as BaseDownloadAssetOptions } from "../../schemas/index";
+import { type DownloadAssetOptions as BaseDownloadAssetOptions, type RPCOptions } from "../../schemas/index";
 export type DownloadAssetOptions = BaseDownloadAssetOptions;
 /**
  * Downloads an asset (model file) without loading it into memory.
@@ -11,6 +11,7 @@ export type DownloadAssetOptions = BaseDownloadAssetOptions;
  *   - assetSrc: The location from which the asset is downloaded (local path, remote URL, or Hyperdrive URL)
  *   - seed: Optional boolean for hyperdrive seeding
  *   - onProgress: Optional callback for download progress
+ * @param rpcOptions - Optional RPC options including per-call profiling configuration
  *
  * @returns Promise that resolves to the asset ID (either the provided assetSrc or a generated ID)
  *
@@ -35,5 +36,5 @@ export type DownloadAssetOptions = BaseDownloadAssetOptions;
  * });
  * ```
  */
-export declare function downloadAsset(options: DownloadAssetOptions): Promise<string>;
+export declare function downloadAsset(options: DownloadAssetOptions, rpcOptions?: RPCOptions): Promise<string>;
 //# sourceMappingURL=download-asset.d.ts.map

package/dist/client/api/download-asset.js

@@ -12,6 +12,7 @@ import { DownloadAssetFailedError, StreamEndedError, InvalidResponseError, } fro
  *   - assetSrc: The location from which the asset is downloaded (local path, remote URL, or Hyperdrive URL)
  *   - seed: Optional boolean for hyperdrive seeding
  *   - onProgress: Optional callback for download progress
+ * @param rpcOptions - Optional RPC options including per-call profiling configuration
  *
  * @returns Promise that resolves to the asset ID (either the provided assetSrc or a generated ID)
  *
@@ -36,11 +37,11 @@ import { DownloadAssetFailedError, StreamEndedError, InvalidResponseError, } fro
  * });
  * ```
  */
-export async function downloadAsset(options) {
+export async function downloadAsset(options, rpcOptions) {
     const request = downloadAssetOptionsToRequestSchema.parse(options);
     if (options.onProgress) {
         // Use streaming for progress updates
-        for await (const response of stream(request)) {
+        for await (const response of stream(request, rpcOptions)) {
             if (response.type === "modelProgress") {
                 options.onProgress(response);
             }
@@ -55,7 +56,7 @@ export async function downloadAsset(options) {
     }
     else {
         // Use regular send for simple downloading
-        const response = await send(request);
+        const response = await send(request, rpcOptions);
         if (response.type !== "downloadAsset") {
             throw new InvalidResponseError("downloadAsset");
         }

package/dist/client/api/embed.d.ts

@@ -1,25 +1,28 @@
+import { type RPCOptions } from "../../schemas/index";
 /**
  * Generates embeddings for a single text using a specified model.
  *
  * @param params - The parameters for the embedding
  * @param params.modelId - The identifier of the embedding model to use
  * @param params.text - The input text to embed
+ * @param options - Optional RPC options including per-call profiling
  * @throws {QvacErrorBase} When the response type is invalid or when the embedding fails
  */
 export declare function embed(params: {
     modelId: string;
     text: string;
-}): Promise<number[]>;
+}, options?: RPCOptions): Promise<number[]>;
 /**
  * Generates embeddings for multiple texts using a specified model.
  *
  * @param params - The parameters for the embedding
  * @param params.modelId - The identifier of the embedding model to use
  * @param params.text - The input texts to embed
+ * @param options - Optional RPC options including per-call profiling
  * @throws {QvacErrorBase} When the response type is invalid or when the embedding fails
  */
 export declare function embed(params: {
     modelId: string;
     text: string[];
-}): Promise<number[][]>;
+}, options?: RPCOptions): Promise<number[][]>;
 //# sourceMappingURL=embed.d.ts.map

package/dist/client/api/embed.js

@@ -1,12 +1,12 @@
 import { send } from "../rpc/rpc-client.js";
 import {} from "../../schemas/index.js";
 import { InvalidResponseError } from "../../utils/errors-client.js";
-export async function embed(params) {
+export async function embed(params, options) {
     const request = {
         type: "embed",
         ...params,
     };
-    const response = await send(request);
+    const response = await send(request, options);
     if (response.type !== "embed") {
         throw new InvalidResponseError("embed");
     }

package/dist/client/api/get-model-info.d.ts

@@ -4,7 +4,7 @@ export declare function getModelInfo(params: GetModelInfoParams): Promise<{
     modelId: string;
     expectedSize: number;
     sha256Checksum: string;
-    addon: "llm" | "whisper" | "embeddings" | "nmt" | "tts" | "ocr" | "vad" | "other";
+    addon: "llm" | "whisper" | "embeddings" | "nmt" | "parakeet" | "tts" | "ocr" | "vad" | "other";
     isCached: boolean;
     isLoaded: boolean;
     cacheFiles: {

package/dist/client/api/heartbeat.d.ts

@@ -0,0 +1,30 @@
+import { type HeartbeatResponse } from "../../schemas/index";
+import type { DelegateBase } from "../../schemas/delegate";
+/**
+ * Checks if a delegated provider is online by sending a heartbeat round-trip.
+ * Can also be used to check if the local SDK worker is responsive.
+ *
+ * @param params - Delegation target to check
+ * @param params.delegate - The provider to check (topic + publicKey + optional timeout)
+ * @returns A promise that resolves to a heartbeat response if the provider is reachable.
+ * @throws {QvacErrorBase} When the provider is unreachable or the response is invalid.
+ *
+ * @example
+ * // Check if a delegated provider is online
+ * try {
+ *   await heartbeat({
+ *     delegate: { topic: "topicHex", providerPublicKey: "peerHex", timeout: 3000 },
+ *   });
+ *   console.log("Provider is online");
+ * } catch {
+ *   console.log("Provider is offline");
+ * }
+ *
+ * @example
+ * // Check if the local SDK worker is responsive
+ * await heartbeat();
+ */
+export declare function heartbeat(params?: {
+    delegate?: DelegateBase;
+}): Promise<HeartbeatResponse>;
+//# sourceMappingURL=heartbeat.d.ts.map

package/dist/client/api/heartbeat.js

@@ -0,0 +1,38 @@
+import {} from "../../schemas/index.js";
+import { send } from "../rpc/rpc-client.js";
+import { InvalidResponseError } from "../../utils/errors-client.js";
+/**
+ * Checks if a delegated provider is online by sending a heartbeat round-trip.
+ * Can also be used to check if the local SDK worker is responsive.
+ *
+ * @param params - Delegation target to check
+ * @param params.delegate - The provider to check (topic + publicKey + optional timeout)
+ * @returns A promise that resolves to a heartbeat response if the provider is reachable.
+ * @throws {QvacErrorBase} When the provider is unreachable or the response is invalid.
+ *
+ * @example
+ * // Check if a delegated provider is online
+ * try {
+ *   await heartbeat({
+ *     delegate: { topic: "topicHex", providerPublicKey: "peerHex", timeout: 3000 },
+ *   });
+ *   console.log("Provider is online");
+ * } catch {
+ *   console.log("Provider is offline");
+ * }
+ *
+ * @example
+ * // Check if the local SDK worker is responsive
+ * await heartbeat();
+ */
+export async function heartbeat(params) {
+    const request = {
+        type: "heartbeat",
+        ...(params?.delegate && { delegate: params.delegate }),
+    };
+    const response = await send(request);
+    if (response.type !== "heartbeat") {
+        throw new InvalidResponseError("heartbeat");
+    }
+    return response;
+}

package/dist/client/api/index.d.ts

@@ -4,7 +4,7 @@ export { completion } from "./completion-stream";
 export { deleteCache } from "./delete-cache";
 export { unloadModel } from "./unload-model";
 export { loggingStream } from "./logging-stream";
-export { ping } from "./ping";
+export { heartbeat } from "./heartbeat";
 export { transcribe, transcribeStream } from "./transcribe";
 export { embed } from "./embed";
 export { translate } from "./translate";

package/dist/client/api/index.js

@@ -4,7 +4,7 @@ export { completion } from "./completion-stream.js";
 export { deleteCache } from "./delete-cache.js";
 export { unloadModel } from "./unload-model.js";
 export { loggingStream } from "./logging-stream.js";
-export { ping } from "./ping.js";
+export { heartbeat } from "./heartbeat.js";
 export { transcribe, transcribeStream } from "./transcribe.js";
 export { embed } from "./embed.js";
 export { translate } from "./translate.js";

package/dist/client/api/invoke-plugin.d.ts

@@ -1,3 +1,4 @@
+import type { RPCOptions } from "../../schemas/index";
 export interface InvokePluginOptions<TParams = unknown> {
     modelId: string;
     handler: string;
@@ -6,9 +7,9 @@ export interface InvokePluginOptions<TParams = unknown> {
 /**
  * Invoke a non-streaming plugin handler.
  */
-export declare function invokePlugin<TResponse = unknown, TParams = unknown>(options: InvokePluginOptions<TParams>): Promise<TResponse>;
+export declare function invokePlugin<TResponse = unknown, TParams = unknown>(options: InvokePluginOptions<TParams>, rpcOptions?: RPCOptions): Promise<TResponse>;
 /**
  * Invoke a streaming plugin handler.
  */
-export declare function invokePluginStream<TResponse = unknown, TParams = unknown>(options: InvokePluginOptions<TParams>): AsyncGenerator<TResponse>;
+export declare function invokePluginStream<TResponse = unknown, TParams = unknown>(options: InvokePluginOptions<TParams>, rpcOptions?: RPCOptions): AsyncGenerator<TResponse>;
 //# sourceMappingURL=invoke-plugin.d.ts.map

package/dist/client/api/invoke-plugin.js

@@ -3,14 +3,14 @@ import { InvalidResponseError } from "../../utils/errors-client.js";
 /**
  * Invoke a non-streaming plugin handler.
  */
-export async function invokePlugin(options) {
+export async function invokePlugin(options, rpcOptions) {
     const request = {
         type: "pluginInvoke",
         modelId: options.modelId,
         handler: options.handler,
         params: options.params,
     };
-    const response = await send(request);
+    const response = await send(request, rpcOptions);
     if (response.type !== "pluginInvoke") {
         throw new InvalidResponseError("pluginInvoke");
     }
@@ -19,14 +19,14 @@ export async function invokePlugin(options) {
 /**
  * Invoke a streaming plugin handler.
  */
-export async function* invokePluginStream(options) {
+export async function* invokePluginStream(options, rpcOptions) {
     const request = {
         type: "pluginInvokeStream",
         modelId: options.modelId,
         handler: options.handler,
         params: options.params,
     };
-    for await (const chunk of stream(request)) {
+    for await (const chunk of stream(request, rpcOptions)) {
         const response = chunk;
         if (response.type !== "pluginInvokeStream") {
             throw new InvalidResponseError("pluginInvokeStream");

package/dist/client/api/load-model.d.ts

@@ -1,4 +1,4 @@
-import { type LoadModelOptions, type ReloadConfigOptions } from "../../schemas/index";
+import { type LoadModelOptions, type ReloadConfigOptions, type RPCOptions } from "../../schemas/index";
 /**
  * Loads a machine learning model from a local path, remote URL, or Hyperdrive key.
  *
@@ -11,11 +11,10 @@ import { type LoadModelOptions, type ReloadConfigOptions } from "../../schemas/i
  * @param options - An object that defines all configuration parameters required for loading the model, including:
  *   - modelSrc: The location from which the model weights are fetched (local path, remote URL, or Hyperdrive URL)
  *   - modelType: The type of model ("llm", "whisper", "embeddings", "nmt", or "tts")
- *   - modelConfig: Model-specific configuration options
- *   - projectionModelSrc: (LLM only) Projection model source for multimodal models
- *   - vadModelSrc: (Whisper only) VAD model source for voice activity detection
+ *   - modelConfig: Model-specific configuration options (companion sources, model parameters, etc.)
  *   - onProgress: Callback for download progress updates
  *   - logger: Logger instance for model operation logs
+ * @param rpcOptions - Optional RPC options including per-call profiling configuration
  *
  * @returns Promise that resolves to the model ID (either the provided modelSrc or a generated ID)
  *
@@ -58,8 +57,10 @@ import { type LoadModelOptions, type ReloadConfigOptions } from "../../schemas/i
  * const multimodalId = await loadModel({
  *   modelSrc: "https://huggingface.co/.../main-model.gguf",
  *   modelType: "llm",
- *   projectionModelSrc: "https://huggingface.co/.../projection-model.gguf",
- *   modelConfig: { ctx_size: 512 },
+ *   modelConfig: {
+ *     ctx_size: 512,
+ *     projectionModelSrc: "https://huggingface.co/.../projection-model.gguf"
+ *   },
  *   onProgress: (progress) => {
  *     console.log(`Loading: ${progress.percentage}%`);
  *   }
@@ -69,12 +70,12 @@ import { type LoadModelOptions, type ReloadConfigOptions } from "../../schemas/i
  * const whisperId = await loadModel({
  *   modelSrc: "https://huggingface.co/.../whisper-model.gguf",
  *   modelType: "whisper",
- *   vadModelSrc: "https://huggingface.co/.../vad-model.bin",
  *   modelConfig: {
  *     mode: "caption",
  *     output_format: "plaintext",
  *     min_seconds: 2,
- *     max_seconds: 6
+ *     max_seconds: 6,
+ *     vadModelSrc: "https://huggingface.co/.../vad-model.bin"
  *   }
  * });
  *
@@ -89,7 +90,7 @@ import { type LoadModelOptions, type ReloadConfigOptions } from "../../schemas/i
  * });
  * ```
  */
-export declare function loadModel(options: LoadModelOptions): Promise<string>;
+export declare function loadModel(options: LoadModelOptions, rpcOptions?: RPCOptions): Promise<string>;
 /**
  * Hot-reloads configuration on an already loaded model.
  *
@@ -97,6 +98,7 @@ export declare function loadModel(options: LoadModelOptions): Promise<string>;
  *   - modelId: The ID of an existing loaded model
  *   - modelType: The type of model (must match the loaded model)
  *   - modelConfig: New configuration to apply
+ * @param rpcOptions - Optional RPC options including per-call profiling configuration
  *
  * @returns Promise that resolves to the model ID
  *
@@ -120,5 +122,5 @@ export declare function loadModel(options: LoadModelOptions): Promise<string>;
  * });
  * ```
  */
-export declare function loadModel(options: ReloadConfigOptions): Promise<string>;
+export declare function loadModel(options: ReloadConfigOptions, rpcOptions?: RPCOptions): Promise<string>;
 //# sourceMappingURL=load-model.d.ts.map

package/dist/client/api/load-model.js

@@ -4,7 +4,7 @@ import { loadModelOptionsToRequestSchema, reloadConfigOptionsToRequestSchema, is
 import { ModelLoadFailedError, StreamEndedError, InvalidResponseError, } from "../../utils/errors-client.js";
 import { getClientLogger } from "../../logging/index.js";
 const logger = getClientLogger();
-export async function loadModel(options) {
+export async function loadModel(options, rpcOptions) {
     const isReloadConfig = "modelId" in options && !("modelSrc" in options);
     // Warn about deprecated alias usage
     if (!isReloadConfig && isModelTypeAlias(options.modelType)) {
@@ -18,7 +18,7 @@ export async function loadModel(options) {
     const onProgress = isReloadConfig ? undefined : options.onProgress;
     if (onProgress) {
         // Use streaming for progress updates
-        for await (const response of stream(request)) {
+        for await (const response of stream(request, rpcOptions)) {
             if (response.type === "modelProgress") {
                 onProgress(response);
             }
@@ -42,7 +42,7 @@ export async function loadModel(options) {
         throw new StreamEndedError();
     }
     // Use regular send for simple loading
-    const response = await send(request);
+    const response = await send(request, rpcOptions);
     if (response.type !== "loadModel") {
         throw new InvalidResponseError("loadModel");
     }