@ai-sdk/openai 1.3.9 → 1.3.11

package/CHANGELOG.md

@@ -1,5 +1,21 @@
 # @ai-sdk/openai
 
+## 1.3.11
+
+### Patch Changes
+
+- beef951: feat: add speech with experimental_generateSpeech
+- Updated dependencies [beef951]
+  - @ai-sdk/provider@1.1.3
+  - @ai-sdk/provider-utils@2.2.7
+
+## 1.3.10
+
+### Patch Changes
+
+- dbe53e7: adding support for gpt-4o-search-preview and handling unsupported parameters
+- 84ffaba: fix: propagate openai transcription fixes
+
 ## 1.3.9
 
 ### Patch Changes

package/dist/index.d.mts

@@ -1,8 +1,8 @@
-import { LanguageModelV1, ProviderV1, EmbeddingModelV1, ImageModelV1, TranscriptionModelV1 } from '@ai-sdk/provider';
+import { LanguageModelV1, ProviderV1, EmbeddingModelV1, ImageModelV1, TranscriptionModelV1, SpeechModelV1 } from '@ai-sdk/provider';
 import { FetchFunction } from '@ai-sdk/provider-utils';
 import { z } from 'zod';
 
-type OpenAIChatModelId = 'o1' | 'o1-2024-12-17' | 'o1-mini' | 'o1-mini-2024-09-12' | 'o1-preview' | 'o1-preview-2024-09-12' | 'o3-mini' | 'o3-mini-2025-01-31' | 'gpt-4o' | 'gpt-4o-2024-05-13' | 'gpt-4o-2024-08-06' | 'gpt-4o-2024-11-20' | 'gpt-4o-audio-preview' | 'gpt-4o-audio-preview-2024-10-01' | 'gpt-4o-audio-preview-2024-12-17' | 'gpt-4o-mini' | 'gpt-4o-mini-2024-07-18' | 'gpt-4-turbo' | 'gpt-4-turbo-2024-04-09' | 'gpt-4-turbo-preview' | 'gpt-4-0125-preview' | 'gpt-4-1106-preview' | 'gpt-4' | 'gpt-4-0613' | 'gpt-4.5-preview' | 'gpt-4.5-preview-2025-02-27' | 'gpt-3.5-turbo-0125' | 'gpt-3.5-turbo' | 'gpt-3.5-turbo-1106' | 'chatgpt-4o-latest' | (string & {});
+type OpenAIChatModelId = 'o1' | 'o1-2024-12-17' | 'o1-mini' | 'o1-mini-2024-09-12' | 'o1-preview' | 'o1-preview-2024-09-12' | 'o3-mini' | 'o3-mini-2025-01-31' | 'gpt-4o' | 'gpt-4o-2024-05-13' | 'gpt-4o-2024-08-06' | 'gpt-4o-2024-11-20' | 'gpt-4o-audio-preview' | 'gpt-4o-audio-preview-2024-10-01' | 'gpt-4o-audio-preview-2024-12-17' | 'gpt-4o-search-preview' | 'gpt-4o-search-preview-2025-03-11' | 'gpt-4o-mini' | 'gpt-4o-mini-2024-07-18' | 'gpt-4-turbo' | 'gpt-4-turbo-2024-04-09' | 'gpt-4-turbo-preview' | 'gpt-4-0125-preview' | 'gpt-4-1106-preview' | 'gpt-4' | 'gpt-4-0613' | 'gpt-4.5-preview' | 'gpt-4.5-preview-2025-02-27' | 'gpt-3.5-turbo-0125' | 'gpt-3.5-turbo' | 'gpt-3.5-turbo-1106' | 'chatgpt-4o-latest' | (string & {});
 interface OpenAIChatSettings {
     /**
   Modify the likelihood of specified tokens appearing in the completion.
@@ -203,6 +203,8 @@ declare const openaiTools: {
     webSearchPreview: typeof webSearchPreviewTool;
 };
 
+type OpenAISpeechModelId = 'tts-1' | 'tts-1-hd' | 'gpt-4o-mini-tts' | (string & {});
+
 interface OpenAIProvider extends ProviderV1 {
     (modelId: 'gpt-3.5-turbo-instruct', settings?: OpenAICompletionSettings): OpenAICompletionLanguageModel;
     (modelId: OpenAIChatModelId, settings?: OpenAIChatSettings): LanguageModelV1;
@@ -250,6 +252,10 @@ interface OpenAIProvider extends ProviderV1 {
      */
     transcription(modelId: OpenAITranscriptionModelId): TranscriptionModelV1;
     /**
+  Creates a model for speech generation.
+     */
+    speech(modelId: OpenAISpeechModelId): SpeechModelV1;
+    /**
   OpenAI-specific tools.
      */
     tools: typeof openaiTools;

package/dist/index.d.ts

@@ -1,8 +1,8 @@
-import { LanguageModelV1, ProviderV1, EmbeddingModelV1, ImageModelV1, TranscriptionModelV1 } from '@ai-sdk/provider';
+import { LanguageModelV1, ProviderV1, EmbeddingModelV1, ImageModelV1, TranscriptionModelV1, SpeechModelV1 } from '@ai-sdk/provider';
 import { FetchFunction } from '@ai-sdk/provider-utils';
 import { z } from 'zod';
 
-type OpenAIChatModelId = 'o1' | 'o1-2024-12-17' | 'o1-mini' | 'o1-mini-2024-09-12' | 'o1-preview' | 'o1-preview-2024-09-12' | 'o3-mini' | 'o3-mini-2025-01-31' | 'gpt-4o' | 'gpt-4o-2024-05-13' | 'gpt-4o-2024-08-06' | 'gpt-4o-2024-11-20' | 'gpt-4o-audio-preview' | 'gpt-4o-audio-preview-2024-10-01' | 'gpt-4o-audio-preview-2024-12-17' | 'gpt-4o-mini' | 'gpt-4o-mini-2024-07-18' | 'gpt-4-turbo' | 'gpt-4-turbo-2024-04-09' | 'gpt-4-turbo-preview' | 'gpt-4-0125-preview' | 'gpt-4-1106-preview' | 'gpt-4' | 'gpt-4-0613' | 'gpt-4.5-preview' | 'gpt-4.5-preview-2025-02-27' | 'gpt-3.5-turbo-0125' | 'gpt-3.5-turbo' | 'gpt-3.5-turbo-1106' | 'chatgpt-4o-latest' | (string & {});
+type OpenAIChatModelId = 'o1' | 'o1-2024-12-17' | 'o1-mini' | 'o1-mini-2024-09-12' | 'o1-preview' | 'o1-preview-2024-09-12' | 'o3-mini' | 'o3-mini-2025-01-31' | 'gpt-4o' | 'gpt-4o-2024-05-13' | 'gpt-4o-2024-08-06' | 'gpt-4o-2024-11-20' | 'gpt-4o-audio-preview' | 'gpt-4o-audio-preview-2024-10-01' | 'gpt-4o-audio-preview-2024-12-17' | 'gpt-4o-search-preview' | 'gpt-4o-search-preview-2025-03-11' | 'gpt-4o-mini' | 'gpt-4o-mini-2024-07-18' | 'gpt-4-turbo' | 'gpt-4-turbo-2024-04-09' | 'gpt-4-turbo-preview' | 'gpt-4-0125-preview' | 'gpt-4-1106-preview' | 'gpt-4' | 'gpt-4-0613' | 'gpt-4.5-preview' | 'gpt-4.5-preview-2025-02-27' | 'gpt-3.5-turbo-0125' | 'gpt-3.5-turbo' | 'gpt-3.5-turbo-1106' | 'chatgpt-4o-latest' | (string & {});
 interface OpenAIChatSettings {
     /**
   Modify the likelihood of specified tokens appearing in the completion.
@@ -203,6 +203,8 @@ declare const openaiTools: {
     webSearchPreview: typeof webSearchPreviewTool;
 };
 
+type OpenAISpeechModelId = 'tts-1' | 'tts-1-hd' | 'gpt-4o-mini-tts' | (string & {});
+
 interface OpenAIProvider extends ProviderV1 {
     (modelId: 'gpt-3.5-turbo-instruct', settings?: OpenAICompletionSettings): OpenAICompletionLanguageModel;
     (modelId: OpenAIChatModelId, settings?: OpenAIChatSettings): LanguageModelV1;
@@ -250,6 +252,10 @@ interface OpenAIProvider extends ProviderV1 {
      */
     transcription(modelId: OpenAITranscriptionModelId): TranscriptionModelV1;
     /**
+  Creates a model for speech generation.
+     */
+    speech(modelId: OpenAISpeechModelId): SpeechModelV1;
+    /**
   OpenAI-specific tools.
      */
     tools: typeof openaiTools;

package/dist/index.js

@@ -26,7 +26,7 @@ __export(src_exports, {
 module.exports = __toCommonJS(src_exports);
 
 // src/openai-provider.ts
-var import_provider_utils10 = require("@ai-sdk/provider-utils");
+var import_provider_utils11 = require("@ai-sdk/provider-utils");
 
 // src/openai-chat-language-model.ts
 var import_provider3 = require("@ai-sdk/provider");
@@ -536,6 +536,15 @@ var OpenAIChatLanguageModel = class {
         }
         baseArgs.max_tokens = void 0;
       }
+    } else if (this.modelId.startsWith("gpt-4o-search-preview")) {
+      if (baseArgs.temperature != null) {
+        baseArgs.temperature = void 0;
+        warnings.push({
+          type: "unsupported-setting",
+          setting: "temperature",
+          details: "temperature is not supported for the gpt-4o-search-preview model and has been removed."
+        });
+      }
     }
     switch (type) {
       case "regular": {
@@ -1604,18 +1613,12 @@ var openaiImageResponseSchema = import_zod5.z.object({
 // src/openai-transcription-model.ts
 var import_provider_utils7 = require("@ai-sdk/provider-utils");
 var import_zod6 = require("zod");
-var OpenAIProviderOptionsSchema = import_zod6.z.object({
-  include: import_zod6.z.array(import_zod6.z.string()).optional().describe(
-    "Additional information to include in the transcription response."
-  ),
-  language: import_zod6.z.string().optional().describe("The language of the input audio in ISO-639-1 format."),
-  prompt: import_zod6.z.string().optional().describe(
-    "An optional text to guide the model's style or continue a previous audio segment."
-  ),
-  temperature: import_zod6.z.number().min(0).max(1).optional().default(0).describe("The sampling temperature, between 0 and 1."),
-  timestampGranularities: import_zod6.z.array(import_zod6.z.enum(["word", "segment"])).optional().default(["segment"]).describe(
-    "The timestamp granularities to populate for this transcription."
-  )
+var openAIProviderOptionsSchema = import_zod6.z.object({
+  include: import_zod6.z.array(import_zod6.z.string()).nullish(),
+  language: import_zod6.z.string().nullish(),
+  prompt: import_zod6.z.string().nullish(),
+  temperature: import_zod6.z.number().min(0).max(1).nullish().default(0),
+  timestampGranularities: import_zod6.z.array(import_zod6.z.enum(["word", "segment"])).nullish().default(["segment"])
 });
 var languageMap = {
   afrikaans: "af",
@@ -1690,11 +1693,12 @@ var OpenAITranscriptionModel = class {
     mediaType,
     providerOptions
   }) {
+    var _a, _b, _c, _d, _e;
     const warnings = [];
     const openAIOptions = (0, import_provider_utils7.parseProviderOptions)({
       provider: "openai",
       providerOptions,
-      schema: OpenAIProviderOptionsSchema
+      schema: openAIProviderOptionsSchema
     });
     const formData = new FormData();
     const blob = audio instanceof Uint8Array ? new Blob([audio]) : new Blob([(0, import_provider_utils7.convertBase64ToUint8Array)(audio)]);
@@ -1702,16 +1706,16 @@ var OpenAITranscriptionModel = class {
     formData.append("file", new File([blob], "audio", { type: mediaType }));
     if (openAIOptions) {
       const transcriptionModelOptions = {
-        include: openAIOptions.include,
-        language: openAIOptions.language,
-        prompt: openAIOptions.prompt,
-        temperature: openAIOptions.temperature,
-        timestamp_granularities: openAIOptions.timestampGranularities
+        include: (_a = openAIOptions.include) != null ? _a : void 0,
+        language: (_b = openAIOptions.language) != null ? _b : void 0,
+        prompt: (_c = openAIOptions.prompt) != null ? _c : void 0,
+        temperature: (_d = openAIOptions.temperature) != null ? _d : void 0,
+        timestamp_granularities: (_e = openAIOptions.timestampGranularities) != null ? _e : void 0
       };
       for (const key in transcriptionModelOptions) {
         const value = transcriptionModelOptions[key];
         if (value !== void 0) {
-          formData.append(key, value);
+          formData.append(key, String(value));
         }
       }
     }
@@ -2590,14 +2594,113 @@ var openaiTools = {
   webSearchPreview: webSearchPreviewTool
 };
 
+// src/openai-speech-model.ts
+var import_provider_utils10 = require("@ai-sdk/provider-utils");
+var import_zod9 = require("zod");
+var OpenAIProviderOptionsSchema = import_zod9.z.object({
+  instructions: import_zod9.z.string().nullish(),
+  speed: import_zod9.z.number().min(0.25).max(4).default(1).nullish()
+});
+var OpenAISpeechModel = class {
+  constructor(modelId, config) {
+    this.modelId = modelId;
+    this.config = config;
+    this.specificationVersion = "v1";
+  }
+  get provider() {
+    return this.config.provider;
+  }
+  getArgs({
+    text,
+    voice = "alloy",
+    outputFormat = "mp3",
+    speed,
+    instructions,
+    providerOptions
+  }) {
+    const warnings = [];
+    const openAIOptions = (0, import_provider_utils10.parseProviderOptions)({
+      provider: "openai",
+      providerOptions,
+      schema: OpenAIProviderOptionsSchema
+    });
+    const requestBody = {
+      model: this.modelId,
+      input: text,
+      voice,
+      response_format: "mp3",
+      speed,
+      instructions
+    };
+    if (outputFormat) {
+      if (["mp3", "opus", "aac", "flac", "wav", "pcm"].includes(outputFormat)) {
+        requestBody.response_format = outputFormat;
+      } else {
+        warnings.push({
+          type: "unsupported-setting",
+          setting: "outputFormat",
+          details: `Unsupported output format: ${outputFormat}. Using mp3 instead.`
+        });
+      }
+    }
+    if (openAIOptions) {
+      const speechModelOptions = {};
+      for (const key in speechModelOptions) {
+        const value = speechModelOptions[key];
+        if (value !== void 0) {
+          requestBody[key] = value;
+        }
+      }
+    }
+    return {
+      requestBody,
+      warnings
+    };
+  }
+  async doGenerate(options) {
+    var _a, _b, _c;
+    const currentDate = (_c = (_b = (_a = this.config._internal) == null ? void 0 : _a.currentDate) == null ? void 0 : _b.call(_a)) != null ? _c : /* @__PURE__ */ new Date();
+    const { requestBody, warnings } = this.getArgs(options);
+    const {
+      value: audio,
+      responseHeaders,
+      rawValue: rawResponse
+    } = await (0, import_provider_utils10.postJsonToApi)({
+      url: this.config.url({
+        path: "/audio/speech",
+        modelId: this.modelId
+      }),
+      headers: (0, import_provider_utils10.combineHeaders)(this.config.headers(), options.headers),
+      body: requestBody,
+      failedResponseHandler: openaiFailedResponseHandler,
+      successfulResponseHandler: (0, import_provider_utils10.createBinaryResponseHandler)(),
+      abortSignal: options.abortSignal,
+      fetch: this.config.fetch
+    });
+    return {
+      audio,
+      warnings,
+      request: {
+        body: JSON.stringify(requestBody)
+      },
+      response: {
+        timestamp: currentDate,
+        modelId: this.modelId,
+        headers: responseHeaders,
+        body: rawResponse
+      }
+    };
+  }
+};
+
 // src/openai-provider.ts
 function createOpenAI(options = {}) {
   var _a, _b, _c;
-  const baseURL = (_a = (0, import_provider_utils10.withoutTrailingSlash)(options.baseURL)) != null ? _a : "https://api.openai.com/v1";
+  const baseURL = (_a = (0, import_provider_utils11.withoutTrailingSlash)(options.baseURL)) != null ? _a : "https://api.openai.com/v1";
   const compatibility = (_b = options.compatibility) != null ? _b : "compatible";
   const providerName = (_c = options.name) != null ? _c : "openai";
   const getHeaders = () => ({
-    Authorization: `Bearer ${(0, import_provider_utils10.loadApiKey)({
+    Authorization: `Bearer ${(0, import_provider_utils11.loadApiKey)({
       apiKey: options.apiKey,
       environmentVariableName: "OPENAI_API_KEY",
       description: "OpenAI"
@@ -2638,6 +2741,12 @@ function createOpenAI(options = {}) {
     headers: getHeaders,
     fetch: options.fetch
   });
+  const createSpeechModel = (modelId) => new OpenAISpeechModel(modelId, {
+    provider: `${providerName}.speech`,
+    url: ({ path }) => `${baseURL}${path}`,
+    headers: getHeaders,
+    fetch: options.fetch
+  });
   const createLanguageModel = (modelId, settings) => {
     if (new.target) {
       throw new Error(
@@ -2674,6 +2783,8 @@ function createOpenAI(options = {}) {
   provider.imageModel = createImageModel;
   provider.transcription = createTranscriptionModel;
   provider.transcriptionModel = createTranscriptionModel;
+  provider.speech = createSpeechModel;
+  provider.speechModel = createSpeechModel;
   provider.tools = openaiTools;
   return provider;
 }