modelfusion 0.127.0 → 0.129.0

package/CHANGELOG.md

@@ -1,5 +1,59 @@
 # Changelog
 
+## v0.129.0 - 2024-01-20
+
+### Changed
+
+- **breaking change**: Usage of Node `async_hooks` has been renamed from `node:async_hooks` to `async_hooks` for easier Webpack configuration. To exclude the `async_hooks` from client-side bundling, you can use the following config for Next.js (`next.config.mjs` or `next.config.js`):
+
+  ```js
+  /**
+   * @type {import('next').NextConfig}
+   */
+  const nextConfig = {
+    webpack: (config, { isServer }) => {
+      if (isServer) {
+        return config;
+      }
+
+      config.resolve = config.resolve ?? {};
+      config.resolve.fallback = config.resolve.fallback ?? {};
+
+      // async hooks is not available in the browser:
+      config.resolve.fallback.async_hooks = false;
+
+      return config;
+    },
+  };
+  ```
+
+## v0.128.0 - 2024-01-20
+
+### Changed
+
+- **breaking change**: ModelFusion uses `Uint8Array` instead of `Buffer` for better cross-platform compatibility (see also ["Goodbye, Node.js Buffer"](https://sindresorhus.com/blog/goodbye-nodejs-buffer)). This can lead to breaking changes in your code if you use `Buffer`-specific methods.
+- **breaking change**: Image content in multi-modal instruction and chat inputs (e.g. for GPT Vision) is passed in the `image` property (instead of `base64Image`) and supports both base64 strings and `Uint8Array` inputs:
+
+  ```ts
+  const image = fs.readFileSync(path.join("data", "example-image.png"));
+
+  const textStream = await streamText({
+    model: openai.ChatTextGenerator({
+      model: "gpt-4-vision-preview",
+      maxGenerationTokens: 1000,
+    }),
+
+    prompt: [
+      openai.ChatMessage.user([
+        { type: "text", text: "Describe the image in detail:\n\n" },
+        { type: "image", image, mimeType: "image/png" },
+      ]),
+    ],
+  });
+  ```
+
+- OpenAI-compatible providers with predefined API configurations have a customized provider name that shows up in the events.
+
 ## v0.127.0 - 2024-01-15
 
 ### Changed

package/README.md

@@ -86,14 +86,14 @@ Multi-modal vision models such as GPT 4 Vision can process images as part of the
 import { streamText, openai } from "modelfusion";
 import { readFileSync } from "fs";
 
-const image = readFileSync("./image.png").toString("base64");
+const image = readFileSync("./image.png");
 
 const textStream = await streamText({
   model: openai.ChatTextGenerator({ model: "gpt-4-vision-preview" }),
   prompt: [
     openai.ChatMessage.user([
       { type: "text", text: "Describe the image in detail:" },
-      { type: "image", base64Image: image, mimeType: "image/png" },
+      { type: "image", image, mimeType: "image/png" },
     ]),
   ],
 });
@@ -219,7 +219,7 @@ Synthesize speech (audio) from text. Also called TTS (text-to-speech).
 ```ts
 import { generateSpeech, lmnt } from "modelfusion";
 
-// `speech` is a Buffer with MP3 audio data
+// `speech` is a Uint8Array with MP3 audio data
 const speech = await generateSpeech({
   model: lmnt.SpeechGenerator({
     voice: "034b632b-df71-46c8-b440-86a42ffc3cf3", // Henry
@@ -257,7 +257,7 @@ const speechStream = await streamSpeech({
 });
 
 for await (const part of speechStream) {
-  // each part is a Buffer with MP3 audio data
+  // each part is a Uint8Array with MP3 audio data
 }
 ```
 
@@ -635,17 +635,11 @@ A web chat with an AI assistant, implemented as a Next.js app.
 
 Ask questions about a PDF document and get answers from the document.
 
-### [Image generator (Next.js)](https://github.com/lgrammel/modelfusion/tree/main/examples/image-generator-next-js)
+### [Next.js / ModelFusion Demos](https://github.com/lgrammel/modelfusion/tree/main/examples/nextjs)
 
-> _Next.js app_, _Stability AI image generation_
+> _Next.js app_, _image generation_
 
-Create an 19th century painting image for your input.
-
-### [Voice recording and transcription (Next.js)](https://github.com/lgrammel/modelfusion/tree/main/examples/voice-recording-next-js)
-
-> _Next.js app_, _OpenAI Whisper_
-
-Record audio with push-to-talk and transcribe it using Whisper, implemented as a Next.js app. The app shows a list of the transcriptions.
+Various examples of using ModelFusion with Next.js: Image generation, voice recording & transcription.
 
 ### [Duplex Speech Streaming (using Vite/React & ModelFusion Server/Fastify)](https://github.com/lgrammel/modelfusion/tree/main/examples/speech-streaming-vite-react-fastify)
 

package/core/api/postToApi.cjs

@@ -1,6 +1,7 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.postToApi = exports.postJsonToApi = exports.createAudioMpegResponseHandler = exports.createTextResponseHandler = exports.createJsonResponseHandler = exports.createTextErrorResponseHandler = exports.createJsonErrorResponseHandler = void 0;
+const UInt8Utils_js_1 = require("../../util/UInt8Utils.cjs");
 const parseJSON_js_1 = require("../schema/parseJSON.cjs");
 const ApiCallError_js_1 = require("./ApiCallError.cjs");
 const createJsonErrorResponseHandler = ({ errorSchema, errorToMessage, isRetryable, }) => async ({ response, url, requestBodyValues }) => {
@@ -86,8 +87,7 @@ const createAudioMpegResponseHandler = () => async ({ response, url, requestBody
             requestBodyValues,
         });
     }
-    const arrayBuffer = await response.arrayBuffer();
-    return Buffer.from(arrayBuffer);
+    return (0, UInt8Utils_js_1.toUint8Array)(await response.arrayBuffer());
 };
 exports.createAudioMpegResponseHandler = createAudioMpegResponseHandler;
 const postJsonToApi = async ({ url, headers, body, failedResponseHandler, successfulResponseHandler, abortSignal, }) => (0, exports.postToApi)({

package/core/api/postToApi.d.ts

@@ -1,4 +1,3 @@
-/// <reference types="node" />
 import { Schema } from "../schema/Schema.js";
 import { ApiCallError } from "./ApiCallError.js";
 export type ResponseHandler<T> = (options: {
@@ -16,7 +15,7 @@ export declare const createTextErrorResponseHandler: ({ isRetryable, }?: {
 }) => ResponseHandler<ApiCallError>;
 export declare const createJsonResponseHandler: <T>(responseSchema: Schema<T>) => ResponseHandler<T>;
 export declare const createTextResponseHandler: () => ResponseHandler<string>;
-export declare const createAudioMpegResponseHandler: () => ResponseHandler<Buffer>;
+export declare const createAudioMpegResponseHandler: () => ResponseHandler<Uint8Array>;
 export declare const postJsonToApi: <T>({ url, headers, body, failedResponseHandler, successfulResponseHandler, abortSignal, }: {
     url: string;
     headers?: Record<string, string> | undefined;
@@ -29,7 +28,7 @@ export declare const postToApi: <T>({ url, headers, body, successfulResponseHand
     url: string;
     headers?: Record<string, string> | undefined;
     body: {
-        content: string | FormData | Buffer;
+        content: string | FormData | Uint8Array;
         values: unknown;
     };
     failedResponseHandler: ResponseHandler<Error>;

package/core/api/postToApi.js

@@ -1,3 +1,4 @@
+import { toUint8Array } from "../../util/UInt8Utils.js";
 import { parseJSON, safeParseJSON } from "../schema/parseJSON.js";
 import { ApiCallError } from "./ApiCallError.js";
 export const createJsonErrorResponseHandler = ({ errorSchema, errorToMessage, isRetryable, }) => async ({ response, url, requestBodyValues }) => {
@@ -79,8 +80,7 @@ export const createAudioMpegResponseHandler = () => async ({ response, url, requ
             requestBodyValues,
         });
     }
-    const arrayBuffer = await response.arrayBuffer();
-    return Buffer.from(arrayBuffer);
+    return toUint8Array(await response.arrayBuffer());
 };
 export const postJsonToApi = async ({ url, headers, body, failedResponseHandler, successfulResponseHandler, abortSignal, }) => postToApi({
     url,

package/core/getFunctionCallLogger.cjs

@@ -61,8 +61,8 @@ const detailedObjectObserver = {
                     if (v === undefined) {
                         return [k, undefined];
                     }
-                    else if (v instanceof Buffer) {
-                        return [k, "omitted<Buffer>"];
+                    else if (v instanceof Uint8Array) {
+                        return [k, "omitted<Uint8Array>"];
                     }
                     else if (Array.isArray(v) &&
                         v.length > 20 &&

package/core/getFunctionCallLogger.js

@@ -57,8 +57,8 @@ const detailedObjectObserver = {
                     if (v === undefined) {
                         return [k, undefined];
                     }
-                    else if (v instanceof Buffer) {
-                        return [k, "omitted<Buffer>"];
+                    else if (v instanceof Uint8Array) {
+                        return [k, "omitted<Uint8Array>"];
                     }
                     else if (Array.isArray(v) &&
                         v.length > 20 &&

package/core/getRun.cjs

@@ -34,7 +34,8 @@ async function ensureLoaded() {
     if (!isNode)
         return Promise.resolve();
     if (!runStorage) {
-        const { AsyncLocalStorage } = await Promise.resolve().then(() => __importStar(require("node:async_hooks")));
+        // Note: using "async_hooks" instead of "node:async_hooks" to avoid webpack fallback problems.
+        const { AsyncLocalStorage } = await Promise.resolve().then(() => __importStar(require("async_hooks")));
         runStorage = new AsyncLocalStorage();
     }
     return Promise.resolve();

package/core/getRun.js

@@ -8,7 +8,8 @@ async function ensureLoaded() {
     if (!isNode)
         return Promise.resolve();
     if (!runStorage) {
-        const { AsyncLocalStorage } = await import("node:async_hooks");
+        // Note: using "async_hooks" instead of "node:async_hooks" to avoid webpack fallback problems.
+        const { AsyncLocalStorage } = await import("async_hooks");
         runStorage = new AsyncLocalStorage();
     }
     return Promise.resolve();

package/model-function/generate-image/generateImage.cjs

@@ -1,6 +1,7 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.generateImage = void 0;
+const UInt8Utils_js_1 = require("../../util/UInt8Utils.cjs");
 const executeStandardCall_js_1 = require("../executeStandardCall.cjs");
 async function generateImage({ model, prompt, fullResponse, ...options }) {
     const callResponse = await (0, executeStandardCall_js_1.executeStandardCall)({
@@ -17,7 +18,7 @@ async function generateImage({ model, prompt, fullResponse, ...options }) {
         },
     });
     const imagesBase64 = callResponse.value;
-    const images = imagesBase64.map((imageBase64) => Buffer.from(imageBase64, "base64"));
+    const images = imagesBase64.map(UInt8Utils_js_1.base64ToUint8Array);
     return fullResponse
         ? {
             image: images[0],

package/model-function/generate-image/generateImage.d.ts

@@ -1,4 +1,3 @@
-/// <reference types="node" />
 import { FunctionOptions } from "../../core/FunctionOptions.js";
 import { ModelCallMetadata } from "../ModelCallMetadata.js";
 import { ImageGenerationModel, ImageGenerationModelSettings } from "./ImageGenerationModel.js";
@@ -23,21 +22,21 @@ import { ImageGenerationModel, ImageGenerationModelSettings } from "./ImageGener
  * @param {PROMPT} prompt - The prompt to be used for image generation.
  *
  * @returns {Promise} - Returns a promise that resolves to the generated image.
- * The image is a Buffer containing the image data in PNG format.
+ * The image is a Uint8Array containing the image data in PNG format.
  */
 export declare function generateImage<PROMPT>(args: {
     model: ImageGenerationModel<PROMPT, ImageGenerationModelSettings>;
     prompt: PROMPT;
     fullResponse?: false;
-} & FunctionOptions): Promise<Buffer>;
+} & FunctionOptions): Promise<Uint8Array>;
 export declare function generateImage<PROMPT>(args: {
     model: ImageGenerationModel<PROMPT, ImageGenerationModelSettings>;
     prompt: PROMPT;
     fullResponse: true;
 } & FunctionOptions): Promise<{
-    image: Buffer;
+    image: Uint8Array;
     imageBase64: string;
-    images: Buffer[];
+    images: Uint8Array[];
     imagesBase64: string[];
     rawResponse: unknown;
     metadata: ModelCallMetadata;

package/model-function/generate-image/generateImage.js

@@ -1,3 +1,4 @@
+import { base64ToUint8Array } from "../../util/UInt8Utils.js";
 import { executeStandardCall } from "../executeStandardCall.js";
 export async function generateImage({ model, prompt, fullResponse, ...options }) {
     const callResponse = await executeStandardCall({
@@ -14,7 +15,7 @@ export async function generateImage({ model, prompt, fullResponse, ...options })
         },
     });
     const imagesBase64 = callResponse.value;
-    const images = imagesBase64.map((imageBase64) => Buffer.from(imageBase64, "base64"));
+    const images = imagesBase64.map(base64ToUint8Array);
     return fullResponse
         ? {
             image: images[0],

package/model-function/generate-speech/SpeechGenerationEvent.d.ts

@@ -1,4 +1,3 @@
-/// <reference types="node" />
 import { BaseModelCallFinishedEvent, BaseModelCallStartedEvent } from "../ModelCallEvent.js";
 export interface SpeechGenerationStartedEvent extends BaseModelCallStartedEvent {
     functionType: "generate-speech";
@@ -7,7 +6,7 @@ export interface SpeechGenerationStartedEvent extends BaseModelCallStartedEvent
 export type SpeechGenerationFinishedEventResult = {
     status: "success";
     rawResponse: unknown;
-    value: Buffer;
+    value: Uint8Array;
 } | {
     status: "error";
     error: unknown;

package/model-function/generate-speech/SpeechGenerationModel.d.ts

@@ -1,4 +1,3 @@
-/// <reference types="node" />
 import { FunctionCallOptions } from "../../core/FunctionOptions.js";
 import { Delta } from "../Delta.js";
 import { Model, ModelSettings } from "../Model.js";
@@ -6,10 +5,10 @@ export interface SpeechGenerationModelSettings extends ModelSettings {
 }
 export interface SpeechGenerationModel<SETTINGS extends SpeechGenerationModelSettings = SpeechGenerationModelSettings> extends Model<SETTINGS> {
     /**
-     * Generates an mp3 audio buffer that contains the speech for the given text.
+     * Generates an mp3 audio Uint8Array that contains the speech for the given text.
      */
-    doGenerateSpeechStandard(text: string, options: FunctionCallOptions): PromiseLike<Buffer>;
+    doGenerateSpeechStandard(text: string, options: FunctionCallOptions): PromiseLike<Uint8Array>;
 }
 export interface StreamingSpeechGenerationModel<SETTINGS extends SpeechGenerationModelSettings = SpeechGenerationModelSettings> extends SpeechGenerationModel<SETTINGS> {
-    doGenerateSpeechStreamDuplex(textStream: AsyncIterable<string>, options: FunctionCallOptions): PromiseLike<AsyncIterable<Delta<Buffer>>>;
+    doGenerateSpeechStreamDuplex(textStream: AsyncIterable<string>, options: FunctionCallOptions): PromiseLike<AsyncIterable<Delta<Uint8Array>>>;
 }

package/model-function/generate-speech/generateSpeech.d.ts

@@ -1,4 +1,3 @@
-/// <reference types="node" />
 import { FunctionOptions } from "../../core/FunctionOptions.js";
 import { ModelCallMetadata } from "../ModelCallMetadata.js";
 import { SpeechGenerationModel, SpeechGenerationModelSettings } from "./SpeechGenerationModel.js";
@@ -17,19 +16,19 @@ import { SpeechGenerationModel, SpeechGenerationModelSettings } from "./SpeechGe
  * @param {SpeechGenerationModel<SpeechGenerationModelSettings>} model - The speech generation model.
  * @param {string} text - The text to be converted to speech.
  *
- * @returns {Promise<Buffer>} - A promise that resolves to a buffer containing the synthesized speech.
+ * @returns {Promise<Uint8Array>} - A promise that resolves to a Uint8Array containing the synthesized speech.
  */
 export declare function generateSpeech(args: {
     model: SpeechGenerationModel<SpeechGenerationModelSettings>;
     text: string;
     fullResponse?: false;
-} & FunctionOptions): Promise<Buffer>;
+} & FunctionOptions): Promise<Uint8Array>;
 export declare function generateSpeech(args: {
     model: SpeechGenerationModel<SpeechGenerationModelSettings>;
     text: string;
     fullResponse: true;
 } & FunctionOptions): Promise<{
-    speech: Buffer;
+    speech: Uint8Array;
     rawResponse: unknown;
     metadata: ModelCallMetadata;
 }>;

package/model-function/generate-speech/streamSpeech.d.ts

@@ -1,4 +1,3 @@
-/// <reference types="node" />
 import { FunctionOptions } from "../../core/FunctionOptions.js";
 import { ModelCallMetadata } from "../ModelCallMetadata.js";
 import { SpeechGenerationModelSettings, StreamingSpeechGenerationModel } from "./SpeechGenerationModel.js";
@@ -24,18 +23,18 @@ import { SpeechGenerationModelSettings, StreamingSpeechGenerationModel } from ".
  * @param {AsyncIterable<string> | string} text - The text to be converted to speech. Can be a string or an async iterable of strings.
  * @param {FunctionOptions} [options] - Optional function options.
  *
- * @returns {AsyncIterableResultPromise<Buffer>} An async iterable promise that contains the synthesized speech chunks.
+ * @returns {AsyncIterableResultPromise<Uint8Array>} An async iterable promise that contains the synthesized speech chunks.
  */
 export declare function streamSpeech(args: {
     model: StreamingSpeechGenerationModel<SpeechGenerationModelSettings>;
     text: AsyncIterable<string> | string;
     fullResponse?: false;
-} & FunctionOptions): Promise<AsyncIterable<Buffer>>;
+} & FunctionOptions): Promise<AsyncIterable<Uint8Array>>;
 export declare function streamSpeech(args: {
     model: StreamingSpeechGenerationModel<SpeechGenerationModelSettings>;
     text: AsyncIterable<string> | string;
     fullResponse: true;
 } & FunctionOptions): Promise<{
-    speechStream: AsyncIterable<Buffer>;
+    speechStream: AsyncIterable<Uint8Array>;
     metadata: Omit<ModelCallMetadata, "durationInMs" | "finishTimestamp">;
 }>;

package/model-function/generate-text/prompt-template/ContentPart.cjs

@@ -1,7 +1,15 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.validateContentIsString = void 0;
+exports.validateContentIsString = exports.getImageAsBase64 = void 0;
+const UInt8Utils_js_1 = require("../../../util/UInt8Utils.cjs");
 const InvalidPromptError_js_1 = require("./InvalidPromptError.cjs");
+function getImageAsBase64(image) {
+    if (typeof image === "string") {
+        return image;
+    }
+    return (0, UInt8Utils_js_1.uint8ArrayToBase64)(image);
+}
+exports.getImageAsBase64 = getImageAsBase64;
 function validateContentIsString(content, prompt) {
     if (typeof content !== "string") {
         throw new InvalidPromptError_js_1.InvalidPromptError("Only text prompts are are supported by this prompt template.", prompt);

package/model-function/generate-text/prompt-template/ContentPart.d.ts

@@ -8,9 +8,9 @@ export interface TextPart {
 export interface ImagePart {
     type: "image";
     /**
-     * Base-64 encoded image.
+     * Image data. Can either be a base64-encoded string or a Uint8Array.
      */
-    base64Image: string;
+    image: string | Uint8Array;
     /**
      * Optional mime type of the image.
      */
@@ -27,4 +27,5 @@ export interface ToolResponsePart {
     id: string;
     response: unknown;
 }
+export declare function getImageAsBase64(image: string | Uint8Array): string;
 export declare function validateContentIsString(content: string | unknown, prompt: unknown): string;

package/model-function/generate-text/prompt-template/ContentPart.js

@@ -1,4 +1,11 @@
+import { uint8ArrayToBase64 } from "../../../util/UInt8Utils.js";
 import { InvalidPromptError } from "./InvalidPromptError.js";
+export function getImageAsBase64(image) {
+    if (typeof image === "string") {
+        return image;
+    }
+    return uint8ArrayToBase64(image);
+}
 export function validateContentIsString(content, prompt) {
     if (typeof content !== "string") {
         throw new InvalidPromptError("Only text prompts are are supported by this prompt template.", prompt);

package/model-provider/elevenlabs/ElevenLabsSpeechModel.cjs

@@ -9,6 +9,7 @@ const parseJSON_js_1 = require("../../core/schema/parseJSON.cjs");
 const AbstractModel_js_1 = require("../../model-function/AbstractModel.cjs");
 const AsyncQueue_js_1 = require("../../util/AsyncQueue.cjs");
 const SimpleWebSocket_js_1 = require("../../util/SimpleWebSocket.cjs");
+const UInt8Utils_js_1 = require("../../util/UInt8Utils.cjs");
 const ElevenLabsApiConfiguration_js_1 = require("./ElevenLabsApiConfiguration.cjs");
 const elevenLabsModels = [
     "eleven_multilingual_v2",
@@ -141,7 +142,7 @@ class ElevenLabsSpeechModel extends AbstractModel_js_1.AbstractModel {
             if (!response.isFinal) {
                 queue.push({
                     type: "delta",
-                    deltaValue: Buffer.from(response.audio, "base64"),
+                    deltaValue: (0, UInt8Utils_js_1.base64ToUint8Array)(response.audio),
                 });
             }
         };

package/model-provider/elevenlabs/ElevenLabsSpeechModel.d.ts

@@ -1,4 +1,3 @@
-/// <reference types="node" />
 import { FunctionCallOptions } from "../../core/FunctionOptions.js";
 import { ApiConfiguration } from "../../core/api/ApiConfiguration.js";
 import { AbstractModel } from "../../model-function/AbstractModel.js";
@@ -37,8 +36,8 @@ export declare class ElevenLabsSpeechModel extends AbstractModel<ElevenLabsSpeec
     get modelName(): string;
     private callAPI;
     get settingsForEvent(): Partial<ElevenLabsSpeechModelSettings>;
-    doGenerateSpeechStandard(text: string, options: FunctionCallOptions): Promise<Buffer>;
-    doGenerateSpeechStreamDuplex(textStream: AsyncIterable<string>): Promise<AsyncIterable<Delta<Buffer>>>;
+    doGenerateSpeechStandard(text: string, options: FunctionCallOptions): Promise<Uint8Array>;
+    doGenerateSpeechStreamDuplex(textStream: AsyncIterable<string>): Promise<AsyncIterable<Delta<Uint8Array>>>;
     withSettings(additionalSettings: Partial<ElevenLabsSpeechModelSettings>): this;
 }
 export {};

package/model-provider/elevenlabs/ElevenLabsSpeechModel.js

@@ -6,6 +6,7 @@ import { safeParseJSON } from "../../core/schema/parseJSON.js";
 import { AbstractModel } from "../../model-function/AbstractModel.js";
 import { AsyncQueue } from "../../util/AsyncQueue.js";
 import { createSimpleWebSocket } from "../../util/SimpleWebSocket.js";
+import { base64ToUint8Array } from "../../util/UInt8Utils.js";
 import { ElevenLabsApiConfiguration } from "./ElevenLabsApiConfiguration.js";
 const elevenLabsModels = [
     "eleven_multilingual_v2",
@@ -138,7 +139,7 @@ export class ElevenLabsSpeechModel extends AbstractModel {
             if (!response.isFinal) {
                 queue.push({
                     type: "delta",
-                    deltaValue: Buffer.from(response.audio, "base64"),
+                    deltaValue: base64ToUint8Array(response.audio),
                 });
             }
         };

package/model-provider/llamacpp/LlamaCppBakLLaVA1PromptTemplate.cjs

@@ -46,7 +46,7 @@ function instruction() {
                         }
                         case "image": {
                             text += `[img-${imageCounter}]`;
-                            images[imageCounter.toString()] = content.base64Image;
+                            images[imageCounter.toString()] = (0, ContentPart_js_1.getImageAsBase64)(content.image);
                             imageCounter++;
                             break;
                         }
@@ -85,7 +85,7 @@ function chat() {
                                 }
                                 case "image": {
                                     text += `[img-${imageCounter}]`;
-                                    images[imageCounter.toString()] = part.base64Image;
+                                    images[imageCounter.toString()] = (0, ContentPart_js_1.getImageAsBase64)(part.image);
                                     imageCounter++;
                                     break;
                                 }

package/model-provider/llamacpp/LlamaCppBakLLaVA1PromptTemplate.js

@@ -1,4 +1,4 @@
-import { validateContentIsString } from "../../model-function/generate-text/prompt-template/ContentPart.js";
+import { getImageAsBase64, validateContentIsString, } from "../../model-function/generate-text/prompt-template/ContentPart.js";
 import { InvalidPromptError } from "../../model-function/generate-text/prompt-template/InvalidPromptError.js";
 import { text as vicunaText } from "../../model-function/generate-text/prompt-template/TextPromptTemplate.js";
 // default Vicuna 1 system message
@@ -42,7 +42,7 @@ export function instruction() {
                         }
                         case "image": {
                             text += `[img-${imageCounter}]`;
-                            images[imageCounter.toString()] = content.base64Image;
+                            images[imageCounter.toString()] = getImageAsBase64(content.image);
                             imageCounter++;
                             break;
                         }
@@ -80,7 +80,7 @@ export function chat() {
                                 }
                                 case "image": {
                                     text += `[img-${imageCounter}]`;
-                                    images[imageCounter.toString()] = part.base64Image;
+                                    images[imageCounter.toString()] = getImageAsBase64(part.image);
                                     imageCounter++;
                                     break;
                                 }

package/model-provider/lmnt/LmntSpeechModel.cjs

@@ -6,6 +6,7 @@ const callWithRetryAndThrottle_js_1 = require("../../core/api/callWithRetryAndTh
 const postToApi_js_1 = require("../../core/api/postToApi.cjs");
 const ZodSchema_js_1 = require("../../core/schema/ZodSchema.cjs");
 const AbstractModel_js_1 = require("../../model-function/AbstractModel.cjs");
+const UInt8Utils_js_1 = require("../../util/UInt8Utils.cjs");
 const LmntApiConfiguration_js_1 = require("./LmntApiConfiguration.cjs");
 /**
  * Synthesize speech using the LMNT API.
@@ -81,7 +82,7 @@ class LmntSpeechModel extends AbstractModel_js_1.AbstractModel {
     }
     async doGenerateSpeechStandard(text, options) {
         const response = await this.callAPI(text, options);
-        return Buffer.from(response.audio, "base64");
+        return (0, UInt8Utils_js_1.base64ToUint8Array)(response.audio);
     }
     withSettings(additionalSettings) {
         return new LmntSpeechModel({

package/model-provider/lmnt/LmntSpeechModel.d.ts

@@ -1,4 +1,3 @@
-/// <reference types="node" />
 import { z } from "zod";
 import { FunctionCallOptions } from "../../core/FunctionOptions.js";
 import { ApiConfiguration } from "../../core/api/ApiConfiguration.js";
@@ -34,7 +33,7 @@ export declare class LmntSpeechModel extends AbstractModel<LmntSpeechModelSettin
     get modelName(): string;
     private callAPI;
     get settingsForEvent(): Partial<LmntSpeechModelSettings>;
-    doGenerateSpeechStandard(text: string, options: FunctionCallOptions): Promise<Buffer>;
+    doGenerateSpeechStandard(text: string, options: FunctionCallOptions): Promise<Uint8Array>;
     withSettings(additionalSettings: Partial<LmntSpeechModelSettings>): this;
 }
 declare const lmntSpeechResponseSchema: z.ZodObject<{

package/model-provider/lmnt/LmntSpeechModel.js

@@ -3,6 +3,7 @@ import { callWithRetryAndThrottle } from "../../core/api/callWithRetryAndThrottl
 import { createJsonResponseHandler, createTextErrorResponseHandler, postToApi, } from "../../core/api/postToApi.js";
 import { zodSchema } from "../../core/schema/ZodSchema.js";
 import { AbstractModel } from "../../model-function/AbstractModel.js";
+import { base64ToUint8Array } from "../../util/UInt8Utils.js";
 import { LmntApiConfiguration } from "./LmntApiConfiguration.js";
 /**
  * Synthesize speech using the LMNT API.
@@ -78,7 +79,7 @@ export class LmntSpeechModel extends AbstractModel {
     }
     async doGenerateSpeechStandard(text, options) {
         const response = await this.callAPI(text, options);
-        return Buffer.from(response.audio, "base64");
+        return base64ToUint8Array(response.audio);
     }
     withSettings(additionalSettings) {
         return new LmntSpeechModel({

package/model-provider/ollama/OllamaChatPromptTemplate.cjs

@@ -95,7 +95,7 @@ function extractUserContent(input) {
             content += part.text;
         }
         else {
-            images.push(part.base64Image);
+            images.push((0, ContentPart_js_1.getImageAsBase64)(part.image));
         }
     }
     return { content, images };

package/model-provider/ollama/OllamaChatPromptTemplate.js

@@ -1,4 +1,4 @@
-import { validateContentIsString } from "../../model-function/generate-text/prompt-template/ContentPart.js";
+import { getImageAsBase64, validateContentIsString, } from "../../model-function/generate-text/prompt-template/ContentPart.js";
 import { InvalidPromptError } from "../../model-function/generate-text/prompt-template/InvalidPromptError.js";
 /**
  * OllamaChatPrompt identity chat format.
@@ -88,7 +88,7 @@ function extractUserContent(input) {
             content += part.text;
         }
         else {
-            images.push(part.base64Image);
+            images.push(getImageAsBase64(part.image));
         }
     }
     return { content, images };

package/model-provider/openai/OpenAIChatMessage.cjs

@@ -1,6 +1,7 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.OpenAIChatMessage = void 0;
+const ContentPart_js_1 = require("../../model-function/generate-text/prompt-template/ContentPart.cjs");
 exports.OpenAIChatMessage = {
     /**
      * Creates a system chat message.
@@ -16,15 +17,15 @@ exports.OpenAIChatMessage = {
             role: "user",
             content: typeof content === "string"
                 ? content
-                : content.map((element) => {
-                    switch (element.type) {
+                : content.map((part) => {
+                    switch (part.type) {
                         case "text": {
-                            return { type: "text", text: element.text };
+                            return { type: "text", text: part.text };
                         }
                         case "image": {
                             return {
                                 type: "image_url",
-                                image_url: `data:${element.mimeType ?? "image/jpeg"};base64,${element.base64Image}`,
+                                image_url: `data:${part.mimeType ?? "image/jpeg"};base64,${(0, ContentPart_js_1.getImageAsBase64)(part.image)}`,
                             };
                         }
                     }

package/model-provider/openai/OpenAIChatMessage.js

@@ -1,3 +1,4 @@
+import { getImageAsBase64, } from "../../model-function/generate-text/prompt-template/ContentPart.js";
 export const OpenAIChatMessage = {
     /**
      * Creates a system chat message.
@@ -13,15 +14,15 @@ export const OpenAIChatMessage = {
             role: "user",
             content: typeof content === "string"
                 ? content
-                : content.map((element) => {
-                    switch (element.type) {
+                : content.map((part) => {
+                    switch (part.type) {
                         case "text": {
-                            return { type: "text", text: element.text };
+                            return { type: "text", text: part.text };
                         }
                         case "image": {
                             return {
                                 type: "image_url",
-                                image_url: `data:${element.mimeType ?? "image/jpeg"};base64,${element.base64Image}`,
+                                image_url: `data:${part.mimeType ?? "image/jpeg"};base64,${getImageAsBase64(part.image)}`,
                             };
                         }
                     }

package/model-provider/openai/OpenAISpeechModel.d.ts

@@ -1,4 +1,3 @@
-/// <reference types="node" />
 import { FunctionCallOptions } from "../../core/FunctionOptions.js";
 import { ApiConfiguration } from "../../core/api/ApiConfiguration.js";
 import { AbstractModel } from "../../model-function/AbstractModel.js";
@@ -46,7 +45,7 @@ export declare class OpenAISpeechModel extends AbstractModel<OpenAISpeechModelSe
     get modelName(): "tts-1" | "tts-1-hd";
     private callAPI;
     get settingsForEvent(): Partial<OpenAISpeechModelSettings>;
-    doGenerateSpeechStandard(text: string, options: FunctionCallOptions): Promise<Buffer>;
+    doGenerateSpeechStandard(text: string, options: FunctionCallOptions): Promise<Uint8Array>;
     withSettings(additionalSettings: Partial<OpenAISpeechModelSettings>): this;
 }
 export {};

package/model-provider/openai/OpenAITranscriptionModel.d.ts

@@ -1,4 +1,3 @@
-/// <reference types="node" />
 import { z } from "zod";
 import { FunctionCallOptions } from "../../core/FunctionOptions.js";
 import { ApiConfiguration } from "../../core/api/ApiConfiguration.js";
@@ -43,7 +42,7 @@ export interface OpenAITranscriptionModelSettings extends TranscriptionModelSett
 }
 export type OpenAITranscriptionInput = {
     type: "flac" | "m4a" | "mp3" | "mp4" | "mpeg" | "mpga" | "ogg" | "wav" | "webm";
-    data: Buffer;
+    data: Uint8Array;
 };
 /**
  * Create a transcription model that calls the OpenAI transcription API.

package/model-provider/openai-compatible/FireworksAIApiConfiguration.cjs

@@ -28,6 +28,12 @@ class FireworksAIApiConfiguration extends BaseUrlApiConfiguration_js_1.BaseUrlAp
                 path: "/inference/v1",
             },
         });
+        Object.defineProperty(this, "provider", {
+            enumerable: true,
+            configurable: true,
+            writable: true,
+            value: "openaicompatible-fireworksai"
+        });
     }
 }
 exports.FireworksAIApiConfiguration = FireworksAIApiConfiguration;

package/model-provider/openai-compatible/FireworksAIApiConfiguration.d.ts

@@ -1,4 +1,5 @@
 import { BaseUrlApiConfigurationWithDefaults, PartialBaseUrlPartsApiConfigurationOptions } from "../../core/api/BaseUrlApiConfiguration.js";
+import { OpenAICompatibleApiConfiguration } from "./OpenAICompatibleApiConfiguration.js";
 /**
  * Configuration for the Fireworks.ai API.
  *
@@ -6,8 +7,9 @@ import { BaseUrlApiConfigurationWithDefaults, PartialBaseUrlPartsApiConfiguratio
  *
  * @see https://readme.fireworks.ai/docs/openai-compatibility
  */
-export declare class FireworksAIApiConfiguration extends BaseUrlApiConfigurationWithDefaults {
+export declare class FireworksAIApiConfiguration extends BaseUrlApiConfigurationWithDefaults implements OpenAICompatibleApiConfiguration {
     constructor(settings?: PartialBaseUrlPartsApiConfigurationOptions & {
         apiKey?: string;
     });
+    readonly provider = "openaicompatible-fireworksai";
 }

package/model-provider/openai-compatible/FireworksAIApiConfiguration.js

@@ -25,5 +25,11 @@ export class FireworksAIApiConfiguration extends BaseUrlApiConfigurationWithDefa
                 path: "/inference/v1",
             },
         });
+        Object.defineProperty(this, "provider", {
+            enumerable: true,
+            configurable: true,
+            writable: true,
+            value: "openaicompatible-fireworksai"
+        });
     }
 }

package/model-provider/openai-compatible/OpenAICompatibleApiConfiguration.d.ts

@@ -0,0 +1,5 @@
+import { ApiConfiguration } from "../../core/api/ApiConfiguration.js";
+export type OpenAICompatibleProviderName = `openaicompatible` | `openaicompatible-${string}`;
+export interface OpenAICompatibleApiConfiguration extends ApiConfiguration {
+    provider?: OpenAICompatibleProviderName;
+}

package/model-provider/openai-compatible/OpenAICompatibleChatModel.cjs

@@ -37,7 +37,7 @@ class OpenAICompatibleChatModel extends AbstractOpenAIChatModel_js_1.AbstractOpe
         });
     }
     get provider() {
-        return this.settings.provider ?? "openaicompatible";
+        return (this.settings.provider ?? this.settings.api.provider ?? "openaicompatible");
     }
     get modelName() {
         return this.settings.model;

package/model-provider/openai-compatible/OpenAICompatibleChatModel.d.ts

@@ -1,4 +1,3 @@
-import { ApiConfiguration } from "../../core/api/ApiConfiguration.js";
 import { FlexibleStructureFromTextPromptTemplate, StructureFromTextPromptTemplate } from "../../model-function/generate-structure/StructureFromTextPromptTemplate.js";
 import { StructureFromTextStreamingModel } from "../../model-function/generate-structure/StructureFromTextStreamingModel.js";
 import { PromptTemplateFullTextModel } from "../../model-function/generate-text/PromptTemplateFullTextModel.js";
@@ -7,9 +6,9 @@ import { TextGenerationPromptTemplate } from "../../model-function/generate-text
 import { ToolCallGenerationModel } from "../../tool/generate-tool-call/ToolCallGenerationModel.js";
 import { ToolCallsGenerationModel } from "../../tool/generate-tool-calls/ToolCallsGenerationModel.js";
 import { AbstractOpenAIChatModel, AbstractOpenAIChatSettings, OpenAIChatPrompt } from "../openai/AbstractOpenAIChatModel.js";
-import { OpenAICompatibleProviderName } from "./OpenAICompatibleProviderName.js";
+import { OpenAICompatibleApiConfiguration, OpenAICompatibleProviderName } from "./OpenAICompatibleApiConfiguration.js";
 export interface OpenAICompatibleChatSettings extends AbstractOpenAIChatSettings {
-    api: ApiConfiguration;
+    api: OpenAICompatibleApiConfiguration;
     provider?: OpenAICompatibleProviderName;
 }
 /**

package/model-provider/openai-compatible/OpenAICompatibleChatModel.js

@@ -34,7 +34,7 @@ export class OpenAICompatibleChatModel extends AbstractOpenAIChatModel {
         });
     }
     get provider() {
-        return this.settings.provider ?? "openaicompatible";
+        return (this.settings.provider ?? this.settings.api.provider ?? "openaicompatible");
     }
     get modelName() {
         return this.settings.model;

package/model-provider/openai-compatible/OpenAICompatibleCompletionModel.cjs

@@ -36,7 +36,7 @@ class OpenAICompatibleCompletionModel extends AbstractOpenAICompletionModel_js_1
         });
     }
     get provider() {
-        return this.settings.provider ?? "openaicompatible";
+        return (this.settings.provider ?? this.settings.api.provider ?? "openaicompatible");
     }
     get modelName() {
         return this.settings.model;

package/model-provider/openai-compatible/OpenAICompatibleCompletionModel.d.ts

@@ -1,11 +1,10 @@
-import { ApiConfiguration } from "../../core/api/ApiConfiguration.js";
 import { PromptTemplateTextStreamingModel } from "../../model-function/generate-text/PromptTemplateTextStreamingModel.js";
 import { TextStreamingBaseModel } from "../../model-function/generate-text/TextGenerationModel.js";
 import { TextGenerationPromptTemplate } from "../../model-function/generate-text/TextGenerationPromptTemplate.js";
 import { AbstractOpenAICompletionModel, AbstractOpenAICompletionModelSettings } from "../openai/AbstractOpenAICompletionModel.js";
-import { OpenAICompatibleProviderName } from "./OpenAICompatibleProviderName.js";
+import { OpenAICompatibleApiConfiguration, OpenAICompatibleProviderName } from "./OpenAICompatibleApiConfiguration.js";
 export interface OpenAICompatibleCompletionModelSettings extends AbstractOpenAICompletionModelSettings {
-    api: ApiConfiguration;
+    api: OpenAICompatibleApiConfiguration;
     provider?: OpenAICompatibleProviderName;
 }
 /**

package/model-provider/openai-compatible/OpenAICompatibleCompletionModel.js

@@ -33,7 +33,7 @@ export class OpenAICompatibleCompletionModel extends AbstractOpenAICompletionMod
         });
     }
     get provider() {
-        return this.settings.provider ?? "openaicompatible";
+        return (this.settings.provider ?? this.settings.api.provider ?? "openaicompatible");
     }
     get modelName() {
         return this.settings.model;

package/model-provider/openai-compatible/OpenAICompatibleTextEmbeddingModel.cjs

@@ -7,7 +7,7 @@ class OpenAICompatibleTextEmbeddingModel extends AbstractOpenAITextEmbeddingMode
         super(settings);
     }
     get provider() {
-        return this.settings.provider ?? "openaicompatible";
+        return (this.settings.provider ?? this.settings.api.provider ?? "openaicompatible");
     }
     get modelName() {
         return this.settings.model;

package/model-provider/openai-compatible/OpenAICompatibleTextEmbeddingModel.d.ts

@@ -1,9 +1,8 @@
-import { ApiConfiguration } from "../../core/api/ApiConfiguration.js";
 import { EmbeddingModel } from "../../model-function/embed/EmbeddingModel.js";
 import { AbstractOpenAITextEmbeddingModel, AbstractOpenAITextEmbeddingModelSettings } from "../openai/AbstractOpenAITextEmbeddingModel.js";
-import { OpenAICompatibleProviderName } from "./OpenAICompatibleProviderName.js";
+import { OpenAICompatibleApiConfiguration, OpenAICompatibleProviderName } from "./OpenAICompatibleApiConfiguration.js";
 export interface OpenAICompatibleTextEmbeddingModelSettings extends AbstractOpenAITextEmbeddingModelSettings {
-    api: ApiConfiguration;
+    api: OpenAICompatibleApiConfiguration;
     provider?: OpenAICompatibleProviderName;
     model: string;
     embeddingDimensions?: number;

package/model-provider/openai-compatible/OpenAICompatibleTextEmbeddingModel.js

@@ -4,7 +4,7 @@ export class OpenAICompatibleTextEmbeddingModel extends AbstractOpenAITextEmbedd
         super(settings);
     }
     get provider() {
-        return this.settings.provider ?? "openaicompatible";
+        return (this.settings.provider ?? this.settings.api.provider ?? "openaicompatible");
     }
     get modelName() {
         return this.settings.model;

package/model-provider/openai-compatible/PerplexityApiConfiguration.cjs

@@ -28,6 +28,12 @@ class PerplexityApiConfiguration extends BaseUrlApiConfiguration_js_1.BaseUrlApi
                 path: "",
             },
         });
+        Object.defineProperty(this, "provider", {
+            enumerable: true,
+            configurable: true,
+            writable: true,
+            value: "openaicompatible-perplexity"
+        });
     }
 }
 exports.PerplexityApiConfiguration = PerplexityApiConfiguration;

package/model-provider/openai-compatible/PerplexityApiConfiguration.d.ts

@@ -1,4 +1,5 @@
 import { BaseUrlApiConfigurationWithDefaults, PartialBaseUrlPartsApiConfigurationOptions } from "../../core/api/BaseUrlApiConfiguration.js";
+import { OpenAICompatibleApiConfiguration } from "./OpenAICompatibleApiConfiguration.js";
 /**
  * Configuration for the Perplexity API.
  *
@@ -6,8 +7,9 @@ import { BaseUrlApiConfigurationWithDefaults, PartialBaseUrlPartsApiConfiguratio
  *
  * @see https://docs.perplexity.ai/reference/post_chat_completions
  */
-export declare class PerplexityApiConfiguration extends BaseUrlApiConfigurationWithDefaults {
+export declare class PerplexityApiConfiguration extends BaseUrlApiConfigurationWithDefaults implements OpenAICompatibleApiConfiguration {
     constructor(settings?: PartialBaseUrlPartsApiConfigurationOptions & {
         apiKey?: string;
     });
+    readonly provider = "openaicompatible-perplexity";
 }

package/model-provider/openai-compatible/PerplexityApiConfiguration.js

@@ -25,5 +25,11 @@ export class PerplexityApiConfiguration extends BaseUrlApiConfigurationWithDefau
                 path: "",
             },
         });
+        Object.defineProperty(this, "provider", {
+            enumerable: true,
+            configurable: true,
+            writable: true,
+            value: "openaicompatible-perplexity"
+        });
     }
 }

package/model-provider/openai-compatible/TogetherAIApiConfiguration.cjs

@@ -28,6 +28,12 @@ class TogetherAIApiConfiguration extends BaseUrlApiConfiguration_js_1.BaseUrlApi
                 path: "/v1",
             },
         });
+        Object.defineProperty(this, "provider", {
+            enumerable: true,
+            configurable: true,
+            writable: true,
+            value: "openaicompatible-togetherai"
+        });
     }
 }
 exports.TogetherAIApiConfiguration = TogetherAIApiConfiguration;

package/model-provider/openai-compatible/TogetherAIApiConfiguration.d.ts

@@ -1,4 +1,5 @@
 import { BaseUrlApiConfigurationWithDefaults, PartialBaseUrlPartsApiConfigurationOptions } from "../../core/api/BaseUrlApiConfiguration.js";
+import { OpenAICompatibleApiConfiguration } from "./OpenAICompatibleApiConfiguration.js";
 /**
  * Configuration for the Together.ai API.
  *
@@ -6,8 +7,9 @@ import { BaseUrlApiConfigurationWithDefaults, PartialBaseUrlPartsApiConfiguratio
  *
  * @see https://docs.together.ai/docs/openai-api-compatibility
  */
-export declare class TogetherAIApiConfiguration extends BaseUrlApiConfigurationWithDefaults {
+export declare class TogetherAIApiConfiguration extends BaseUrlApiConfigurationWithDefaults implements OpenAICompatibleApiConfiguration {
     constructor(settings?: PartialBaseUrlPartsApiConfigurationOptions & {
         apiKey?: string;
     });
+    readonly provider = "openaicompatible-togetherai";
 }

package/model-provider/openai-compatible/TogetherAIApiConfiguration.js

@@ -25,5 +25,11 @@ export class TogetherAIApiConfiguration extends BaseUrlApiConfigurationWithDefau
                 path: "/v1",
             },
         });
+        Object.defineProperty(this, "provider", {
+            enumerable: true,
+            configurable: true,
+            writable: true,
+            value: "openaicompatible-togetherai"
+        });
     }
 }

package/model-provider/openai-compatible/index.cjs

@@ -28,10 +28,10 @@ var __importStar = (this && this.__importStar) || function (mod) {
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.openaicompatible = void 0;
 __exportStar(require("./FireworksAIApiConfiguration.cjs"), exports);
+__exportStar(require("./OpenAICompatibleApiConfiguration.cjs"), exports);
 __exportStar(require("./OpenAICompatibleChatModel.cjs"), exports);
 __exportStar(require("./OpenAICompatibleCompletionModel.cjs"), exports);
 exports.openaicompatible = __importStar(require("./OpenAICompatibleFacade.cjs"));
-__exportStar(require("./OpenAICompatibleProviderName.cjs"), exports);
 __exportStar(require("./OpenAICompatibleTextEmbeddingModel.cjs"), exports);
 __exportStar(require("./PerplexityApiConfiguration.cjs"), exports);
 __exportStar(require("./TogetherAIApiConfiguration.cjs"), exports);

package/model-provider/openai-compatible/index.d.ts

@@ -1,8 +1,8 @@
 export * from "./FireworksAIApiConfiguration.js";
+export * from "./OpenAICompatibleApiConfiguration.js";
 export * from "./OpenAICompatibleChatModel.js";
 export * from "./OpenAICompatibleCompletionModel.js";
 export * as openaicompatible from "./OpenAICompatibleFacade.js";
-export * from "./OpenAICompatibleProviderName.js";
 export * from "./OpenAICompatibleTextEmbeddingModel.js";
 export * from "./PerplexityApiConfiguration.js";
 export * from "./TogetherAIApiConfiguration.js";

package/model-provider/openai-compatible/index.js

@@ -1,8 +1,8 @@
 export * from "./FireworksAIApiConfiguration.js";
+export * from "./OpenAICompatibleApiConfiguration.js";
 export * from "./OpenAICompatibleChatModel.js";
 export * from "./OpenAICompatibleCompletionModel.js";
 export * as openaicompatible from "./OpenAICompatibleFacade.js";
-export * from "./OpenAICompatibleProviderName.js";
 export * from "./OpenAICompatibleTextEmbeddingModel.js";
 export * from "./PerplexityApiConfiguration.js";
 export * from "./TogetherAIApiConfiguration.js";

package/model-provider/stability/StabilityImageGenerationModel.d.ts

@@ -69,9 +69,9 @@ export declare class StabilityImageGenerationModel extends AbstractModel<Stabili
     doGenerateImages(prompt: StabilityImageGenerationPrompt, callOptions: FunctionCallOptions): Promise<{
         rawResponse: {
             artifacts: {
-                base64: string;
                 finishReason: "ERROR" | "SUCCESS" | "CONTENT_FILTERED";
                 seed: number;
+                base64: string;
             }[];
         };
         base64Images: string[];
@@ -86,25 +86,25 @@ declare const stabilityImageGenerationResponseSchema: z.ZodObject<{
         seed: z.ZodNumber;
         finishReason: z.ZodEnum<["SUCCESS", "ERROR", "CONTENT_FILTERED"]>;
     }, "strip", z.ZodTypeAny, {
-        base64: string;
         finishReason: "ERROR" | "SUCCESS" | "CONTENT_FILTERED";
         seed: number;
-    }, {
         base64: string;
+    }, {
         finishReason: "ERROR" | "SUCCESS" | "CONTENT_FILTERED";
         seed: number;
+        base64: string;
     }>, "many">;
 }, "strip", z.ZodTypeAny, {
     artifacts: {
-        base64: string;
         finishReason: "ERROR" | "SUCCESS" | "CONTENT_FILTERED";
         seed: number;
+        base64: string;
     }[];
 }, {
     artifacts: {
-        base64: string;
         finishReason: "ERROR" | "SUCCESS" | "CONTENT_FILTERED";
         seed: number;
+        base64: string;
     }[];
 }>;
 export type StabilityImageGenerationResponse = z.infer<typeof stabilityImageGenerationResponseSchema>;

package/model-provider/whispercpp/WhisperCppTranscriptionModel.d.ts

@@ -1,4 +1,3 @@
-/// <reference types="node" />
 import { FunctionCallOptions } from "../../core/FunctionOptions.js";
 import { ApiConfiguration } from "../../core/api/ApiConfiguration.js";
 import { AbstractModel } from "../../model-function/AbstractModel.js";
@@ -9,7 +8,7 @@ export interface WhisperCppTranscriptionModelSettings extends TranscriptionModel
 }
 export type WhisperCppTranscriptionInput = {
     type: "wav";
-    data: Buffer;
+    data: Uint8Array;
 };
 export declare class WhisperCppTranscriptionModel extends AbstractModel<WhisperCppTranscriptionModelSettings> implements TranscriptionModel<WhisperCppTranscriptionInput, WhisperCppTranscriptionModelSettings> {
     constructor(settings: WhisperCppTranscriptionModelSettings);

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "modelfusion",
   "description": "The TypeScript library for building AI applications.",
-  "version": "0.127.0",
+  "version": "0.129.0",
   "author": "Lars Grammel",
   "license": "MIT",
   "keywords": [
@@ -80,7 +80,7 @@
     "@vitest/ui": "1.1.0",
     "eslint": "^8.45.0",
     "eslint-config-prettier": "9.1.0",
-    "msw": "2.0.11",
+    "msw": "2.1.2",
     "type-fest": "4.9.0"
   }
 }

package/util/UInt8Utils.cjs

@@ -0,0 +1,50 @@
+"use strict";
+/*
+ * MIT License
+ *
+ * Copyright (c) Sindre Sorhus <sindresorhus@gmail.com> (https://sindresorhus.com)
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.uint8ArrayToBase64 = exports.base64ToUint8Array = exports.toUint8Array = void 0;
+// copied from: https://github.com/sindresorhus/uint8array-extras/blob/main/index.js
+function toUint8Array(value) {
+    if (value instanceof ArrayBuffer) {
+        return new Uint8Array(value);
+    }
+    if (ArrayBuffer.isView(value)) {
+        return new Uint8Array(value.buffer, value.byteOffset, value.byteLength);
+    }
+    throw new TypeError(`Unsupported value, got \`${typeof value}\`.`);
+}
+exports.toUint8Array = toUint8Array;
+function base64ToUint8Array(base64String) {
+    return Uint8Array.from(globalThis.atob(base64UrlToBase64(base64String)), (x) => x.codePointAt(0));
+}
+exports.base64ToUint8Array = base64ToUint8Array;
+// Reference: https://phuoc.ng/collection/this-vs-that/concat-vs-push/
+const MAX_BLOCK_SIZE = 65535;
+function uint8ArrayToBase64(array) {
+    let base64;
+    if (array.length < MAX_BLOCK_SIZE) {
+        // Required as `btoa` and `atob` don't properly support Unicode: https://developer.mozilla.org/en-US/docs/Glossary/Base64#the_unicode_problem
+        base64 = globalThis.btoa(String.fromCodePoint(...array));
+    }
+    else {
+        base64 = "";
+        for (const value of array) {
+            base64 += String.fromCodePoint(value);
+        }
+        base64 = globalThis.btoa(base64);
+    }
+    return base64;
+}
+exports.uint8ArrayToBase64 = uint8ArrayToBase64;
+function base64UrlToBase64(base64url) {
+    return base64url.replaceAll("-", "+").replaceAll("_", "/");
+}

package/util/UInt8Utils.d.ts

@@ -0,0 +1,3 @@
+export declare function toUint8Array(value: unknown): Uint8Array;
+export declare function base64ToUint8Array(base64String: string): Uint8Array;
+export declare function uint8ArrayToBase64(array: Uint8Array): string;

package/util/UInt8Utils.js

@@ -0,0 +1,44 @@
+/*
+ * MIT License
+ *
+ * Copyright (c) Sindre Sorhus <sindresorhus@gmail.com> (https://sindresorhus.com)
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+// copied from: https://github.com/sindresorhus/uint8array-extras/blob/main/index.js
+export function toUint8Array(value) {
+    if (value instanceof ArrayBuffer) {
+        return new Uint8Array(value);
+    }
+    if (ArrayBuffer.isView(value)) {
+        return new Uint8Array(value.buffer, value.byteOffset, value.byteLength);
+    }
+    throw new TypeError(`Unsupported value, got \`${typeof value}\`.`);
+}
+export function base64ToUint8Array(base64String) {
+    return Uint8Array.from(globalThis.atob(base64UrlToBase64(base64String)), (x) => x.codePointAt(0));
+}
+// Reference: https://phuoc.ng/collection/this-vs-that/concat-vs-push/
+const MAX_BLOCK_SIZE = 65535;
+export function uint8ArrayToBase64(array) {
+    let base64;
+    if (array.length < MAX_BLOCK_SIZE) {
+        // Required as `btoa` and `atob` don't properly support Unicode: https://developer.mozilla.org/en-US/docs/Glossary/Base64#the_unicode_problem
+        base64 = globalThis.btoa(String.fromCodePoint(...array));
+    }
+    else {
+        base64 = "";
+        for (const value of array) {
+            base64 += String.fromCodePoint(value);
+        }
+        base64 = globalThis.btoa(base64);
+    }
+    return base64;
+}
+function base64UrlToBase64(base64url) {
+    return base64url.replaceAll("-", "+").replaceAll("_", "/");
+}

package/model-provider/openai-compatible/OpenAICompatibleProviderName.d.ts

@@ -1 +0,0 @@
-export type OpenAICompatibleProviderName = `openaicompatible` | `openaicompatible-${string}`;