@ai-sdk/baseten 2.0.0-beta.23 → 2.0.0-beta.25

package/CHANGELOG.md

@@ -1,5 +1,28 @@
 # @ai-sdk/baseten
 
+## 2.0.0-beta.25
+
+### Patch Changes
+
+- Updated dependencies [b3976a2]
+- Updated dependencies [ff5eba1]
+  - @ai-sdk/provider-utils@5.0.0-beta.20
+  - @ai-sdk/openai-compatible@3.0.0-beta.25
+  - @ai-sdk/provider@4.0.0-beta.12
+
+## 2.0.0-beta.24
+
+### Major Changes
+
+- ef992f8: Remove CommonJS exports from all packages. All packages are now ESM-only (`"type": "module"`). Consumers using `require()` must switch to ESM `import` syntax.
+
+### Patch Changes
+
+- Updated dependencies [ef992f8]
+  - @ai-sdk/openai-compatible@3.0.0-beta.24
+  - @ai-sdk/provider@4.0.0-beta.11
+  - @ai-sdk/provider-utils@5.0.0-beta.19
+
 ## 2.0.0-beta.23
 
 ### Patch Changes

package/dist/index.js

@@ -1,44 +1,25 @@
-"use strict";
-var __defProp = Object.defineProperty;
-var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
-var __getOwnPropNames = Object.getOwnPropertyNames;
-var __hasOwnProp = Object.prototype.hasOwnProperty;
-var __export = (target, all) => {
-  for (var name in all)
-    __defProp(target, name, { get: all[name], enumerable: true });
-};
-var __copyProps = (to, from, except, desc) => {
-  if (from && typeof from === "object" || typeof from === "function") {
-    for (let key of __getOwnPropNames(from))
-      if (!__hasOwnProp.call(to, key) && key !== except)
-        __defProp(to, key, { get: () => from[key], enumerable: !(desc = __getOwnPropDesc(from, key)) || desc.enumerable });
-  }
-  return to;
-};
-var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
-
-// src/index.ts
-var index_exports = {};
-__export(index_exports, {
-  VERSION: () => VERSION,
-  baseten: () => baseten,
-  createBaseten: () => createBaseten
-});
-module.exports = __toCommonJS(index_exports);
-
 // src/baseten-provider.ts
-var import_openai_compatible = require("@ai-sdk/openai-compatible");
-var import_provider = require("@ai-sdk/provider");
-var import_provider_utils = require("@ai-sdk/provider-utils");
-var import_v4 = require("zod/v4");
-var import_performance_client = require("@basetenlabs/performance-client");
+import {
+  OpenAICompatibleChatLanguageModel,
+  OpenAICompatibleEmbeddingModel
+} from "@ai-sdk/openai-compatible";
+import {
+  NoSuchModelError
+} from "@ai-sdk/provider";
+import {
+  loadApiKey,
+  withoutTrailingSlash,
+  withUserAgentSuffix
+} from "@ai-sdk/provider-utils";
+import { z } from "zod/v4";
+import { PerformanceClient } from "@basetenlabs/performance-client";
 
 // src/version.ts
-var VERSION = true ? "2.0.0-beta.23" : "0.0.0-test";
+var VERSION = true ? "2.0.0-beta.25" : "0.0.0-test";
 
 // src/baseten-provider.ts
-var basetenErrorSchema = import_v4.z.object({
-  error: import_v4.z.string()
+var basetenErrorSchema = z.object({
+  error: z.string()
 });
 var basetenErrorStructure = {
   errorSchema: basetenErrorSchema,
@@ -47,10 +28,10 @@ var basetenErrorStructure = {
 var defaultBaseURL = "https://inference.baseten.co/v1";
 function createBaseten(options = {}) {
   var _a;
-  const baseURL = (0, import_provider_utils.withoutTrailingSlash)((_a = options.baseURL) != null ? _a : defaultBaseURL);
-  const getHeaders = () => (0, import_provider_utils.withUserAgentSuffix)(
+  const baseURL = withoutTrailingSlash((_a = options.baseURL) != null ? _a : defaultBaseURL);
+  const getHeaders = () => withUserAgentSuffix(
     {
-      Authorization: `Bearer ${(0, import_provider_utils.loadApiKey)({
+      Authorization: `Bearer ${loadApiKey({
         apiKey: options.apiKey,
         environmentVariableName: "BASETEN_API_KEY",
         description: "Baseten API key"
@@ -75,7 +56,7 @@ function createBaseten(options = {}) {
     if (customURL) {
       const isOpenAICompatible = customURL.includes("/sync/v1");
       if (isOpenAICompatible) {
-        return new import_openai_compatible.OpenAICompatibleChatLanguageModel(modelId != null ? modelId : "placeholder", {
+        return new OpenAICompatibleChatLanguageModel(modelId != null ? modelId : "placeholder", {
           ...getCommonModelConfig("chat", customURL),
           errorStructure: basetenErrorStructure
         });
@@ -85,7 +66,7 @@ function createBaseten(options = {}) {
         );
       }
     }
-    return new import_openai_compatible.OpenAICompatibleChatLanguageModel(modelId != null ? modelId : "chat", {
+    return new OpenAICompatibleChatLanguageModel(modelId != null ? modelId : "chat", {
       ...getCommonModelConfig("chat"),
       errorStructure: basetenErrorStructure
     });
@@ -99,7 +80,7 @@ function createBaseten(options = {}) {
     }
     const isOpenAICompatible = customURL.includes("/sync");
     if (isOpenAICompatible) {
-      const model = new import_openai_compatible.OpenAICompatibleEmbeddingModel(
+      const model = new OpenAICompatibleEmbeddingModel(
         modelId != null ? modelId : "embeddings",
         {
           ...getCommonModelConfig("embedding", customURL),
@@ -107,9 +88,9 @@ function createBaseten(options = {}) {
         }
       );
       const performanceClientURL = customURL.replace("/sync/v1", "/sync");
-      const performanceClient = new import_performance_client.PerformanceClient(
+      const performanceClient = new PerformanceClient(
         performanceClientURL,
-        (0, import_provider_utils.loadApiKey)({
+        loadApiKey({
           apiKey: options.apiKey,
           environmentVariableName: "BASETEN_API_KEY",
           description: "Baseten API key"
@@ -144,17 +125,16 @@ function createBaseten(options = {}) {
   provider.chatModel = createChatModel;
   provider.languageModel = createChatModel;
   provider.imageModel = (modelId) => {
-    throw new import_provider.NoSuchModelError({ modelId, modelType: "imageModel" });
+    throw new NoSuchModelError({ modelId, modelType: "imageModel" });
   };
   provider.embeddingModel = createEmbeddingModel;
   provider.textEmbeddingModel = createEmbeddingModel;
   return provider;
 }
 var baseten = createBaseten();
-// Annotate the CommonJS export names for ESM import in node:
-0 && (module.exports = {
+export {
   VERSION,
   baseten,
   createBaseten
-});
+};
 //# sourceMappingURL=index.js.map

package/dist/index.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../src/index.ts","../src/baseten-provider.ts","../src/version.ts"],"sourcesContent":["export type { BasetenChatModelId } from './baseten-chat-options';\nexport { baseten, createBaseten } from './baseten-provider';\nexport type {\n  BasetenProvider,\n  BasetenProviderSettings,\n  BasetenErrorData,\n} from './baseten-provider';\nexport type { BasetenEmbeddingModelOptions } from './baseten-embedding-options';\nexport { VERSION } from './version';\n","import {\n  OpenAICompatibleChatLanguageModel,\n  OpenAICompatibleEmbeddingModel,\n  ProviderErrorStructure,\n} from '@ai-sdk/openai-compatible';\nimport {\n  EmbeddingModelV4,\n  LanguageModelV4,\n  NoSuchModelError,\n  ProviderV4,\n} from '@ai-sdk/provider';\nimport {\n  FetchFunction,\n  loadApiKey,\n  withoutTrailingSlash,\n  withUserAgentSuffix,\n} from '@ai-sdk/provider-utils';\nimport { z } from 'zod/v4';\nimport { BasetenChatModelId } from './baseten-chat-options';\nimport { BasetenEmbeddingModelId } from './baseten-embedding-options';\nimport { PerformanceClient } from '@basetenlabs/performance-client';\nimport { VERSION } from './version';\n\nexport type BasetenErrorData = z.infer<typeof basetenErrorSchema>;\n\nconst basetenErrorSchema = z.object({\n  error: z.string(),\n});\n\nconst basetenErrorStructure: ProviderErrorStructure<BasetenErrorData> = {\n  errorSchema: basetenErrorSchema,\n  errorToMessage: data => data.error,\n};\n\nexport interface BasetenProviderSettings {\n  /**\n   * Baseten API key. Default value is taken from the `BASETEN_API_KEY`\n   * environment variable.\n   */\n  apiKey?: string;\n\n  /**\n   * Base URL for the Model APIs. Default: 'https://inference.baseten.co/v1'\n   */\n  baseURL?: string;\n\n  /**\n   * Model URL for custom models (chat or embeddings).\n   * If not supplied, the default Model APIs will be used.\n   */\n  modelURL?: string;\n  /**\n   * Custom headers to include in the requests.\n   */\n  headers?: Record<string, string>;\n\n  /**\n   * Custom fetch implementation. You can use it as a middleware to intercept requests,\n   * or to provide a custom fetch implementation for e.g. testing.\n   */\n  fetch?: FetchFunction;\n}\n\nexport interface BasetenProvider extends ProviderV4 {\n  /**\n   * Creates a chat model for text generation.\n   */\n  (modelId?: BasetenChatModelId): LanguageModelV4;\n\n  /**\n   * Creates a chat model for text generation.\n   */\n  chatModel(modelId?: BasetenChatModelId): LanguageModelV4;\n\n  /**\n   * Creates a language model for text generation. Alias for chatModel.\n   */\n  languageModel(modelId?: BasetenChatModelId): LanguageModelV4;\n\n  /**\n   * Creates a embedding model for text generation.\n   */\n  embeddingModel(modelId?: BasetenEmbeddingModelId): EmbeddingModelV4;\n\n  /**\n   * @deprecated Use `embeddingModel` instead.\n   */\n  textEmbeddingModel(modelId?: BasetenEmbeddingModelId): EmbeddingModelV4;\n}\n\n// by default, we use the Model APIs\nconst defaultBaseURL = 'https://inference.baseten.co/v1';\n\nexport function createBaseten(\n  options: BasetenProviderSettings = {},\n): BasetenProvider {\n  const baseURL = withoutTrailingSlash(options.baseURL ?? defaultBaseURL);\n  const getHeaders = () =>\n    withUserAgentSuffix(\n      {\n        Authorization: `Bearer ${loadApiKey({\n          apiKey: options.apiKey,\n          environmentVariableName: 'BASETEN_API_KEY',\n          description: 'Baseten API key',\n        })}`,\n        ...options.headers,\n      },\n      `ai-sdk/baseten/${VERSION}`,\n    );\n\n  interface CommonModelConfig {\n    provider: string;\n    url: ({ path }: { path: string }) => string;\n    headers: () => Record<string, string>;\n    fetch?: FetchFunction;\n  }\n\n  const getCommonModelConfig = (\n    modelType: string,\n    customURL?: string,\n  ): CommonModelConfig => ({\n    provider: `baseten.${modelType}`,\n    url: ({ path }) => {\n      // For embeddings with /sync URLs (but not /sync/v1), we need to add /v1\n      if (\n        modelType === 'embedding' &&\n        customURL?.includes('/sync') &&\n        !customURL?.includes('/sync/v1')\n      ) {\n        return `${customURL}/v1${path}`;\n      }\n      return `${customURL || baseURL}${path}`;\n    },\n    headers: getHeaders,\n    fetch: options.fetch,\n  });\n\n  const createChatModel = (modelId?: BasetenChatModelId) => {\n    // Use modelURL if provided, otherwise use default Model APIs\n    const customURL = options.modelURL;\n\n    if (customURL) {\n      // Check if this is a /sync/v1 endpoint (OpenAI-compatible) or /predict endpoint (custom)\n      const isOpenAICompatible = customURL.includes('/sync/v1');\n\n      if (isOpenAICompatible) {\n        // For /sync/v1 endpoints, use standard OpenAI-compatible format\n        return new OpenAICompatibleChatLanguageModel(modelId ?? 'placeholder', {\n          ...getCommonModelConfig('chat', customURL),\n          errorStructure: basetenErrorStructure,\n        });\n      } else if (customURL.includes('/predict')) {\n        throw new Error(\n          'Not supported. You must use a /sync/v1 endpoint for chat models.',\n        );\n      }\n    }\n\n    // Use default OpenAI-compatible format for Model APIs\n    return new OpenAICompatibleChatLanguageModel(modelId ?? 'chat', {\n      ...getCommonModelConfig('chat'),\n      errorStructure: basetenErrorStructure,\n    });\n  };\n\n  const createEmbeddingModel = (modelId?: BasetenEmbeddingModelId) => {\n    // Use modelURL if provided\n    const customURL = options.modelURL;\n    if (!customURL) {\n      throw new Error(\n        'No model URL provided for embeddings. Please set modelURL option for embeddings.',\n      );\n    }\n\n    // Check if this is a /sync or /sync/v1 endpoint (OpenAI-compatible)\n    // We support both /sync and /sync/v1, stripping /v1 before passing to Performance Client, as Performance Client adds /v1 itself\n    const isOpenAICompatible = customURL.includes('/sync');\n\n    if (isOpenAICompatible) {\n      // Create the model using OpenAICompatibleEmbeddingModel and override doEmbed\n      const model = new OpenAICompatibleEmbeddingModel(\n        modelId ?? 'embeddings',\n        {\n          ...getCommonModelConfig('embedding', customURL),\n          errorStructure: basetenErrorStructure,\n        },\n      );\n\n      // Strip /v1 from URL if present before passing to Performance Client to avoid double /v1\n      const performanceClientURL = customURL.replace('/sync/v1', '/sync');\n\n      // Initialize the B10 Performance Client once for reuse\n      const performanceClient = new PerformanceClient(\n        performanceClientURL,\n        loadApiKey({\n          apiKey: options.apiKey,\n          environmentVariableName: 'BASETEN_API_KEY',\n          description: 'Baseten API key',\n        }),\n      );\n\n      // Override the doEmbed method to use the pre-created Performance Client\n      model.doEmbed = async params => {\n        if (!params.values || !Array.isArray(params.values)) {\n          throw new Error('params.values must be an array of strings');\n        }\n\n        // Performance Client handles batching internally, so we don't need to limit in 128 here\n        const response = await performanceClient.embed(\n          params.values,\n          modelId ?? 'embeddings', // model_id is for Model APIs, we don't use it here for dedicated\n        );\n        // Transform the response to match the expected format\n        const embeddings = response.data.map((item: any) => item.embedding);\n\n        return {\n          embeddings,\n          usage: response.usage\n            ? { tokens: response.usage.total_tokens }\n            : undefined,\n          response: { headers: {}, body: response },\n          warnings: [],\n        };\n      };\n\n      return model;\n    } else {\n      throw new Error(\n        'Not supported. You must use a /sync or /sync/v1 endpoint for embeddings.',\n      );\n    }\n  };\n\n  const provider = (modelId?: BasetenChatModelId) => createChatModel(modelId);\n\n  provider.specificationVersion = 'v4' as const;\n  provider.chatModel = createChatModel;\n  provider.languageModel = createChatModel;\n  provider.imageModel = (modelId: string) => {\n    throw new NoSuchModelError({ modelId, modelType: 'imageModel' });\n  };\n  provider.embeddingModel = createEmbeddingModel;\n  provider.textEmbeddingModel = createEmbeddingModel;\n  return provider;\n}\n\nexport const baseten = createBaseten();\n","// Version string of this package injected at build time.\ndeclare const __PACKAGE_VERSION__: string | undefined;\nexport const VERSION: string =\n  typeof __PACKAGE_VERSION__ !== 'undefined'\n    ? __PACKAGE_VERSION__\n    : '0.0.0-test';\n"],"mappings":";;;;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;;;ACAA,+BAIO;AACP,sBAKO;AACP,4BAKO;AACP,gBAAkB;AAGlB,gCAAkC;;;AClB3B,IAAM,UACX,OACI,kBACA;;;ADoBN,IAAM,qBAAqB,YAAE,OAAO;AAAA,EAClC,OAAO,YAAE,OAAO;AAClB,CAAC;AAED,IAAM,wBAAkE;AAAA,EACtE,aAAa;AAAA,EACb,gBAAgB,UAAQ,KAAK;AAC/B;AA2DA,IAAM,iBAAiB;AAEhB,SAAS,cACd,UAAmC,CAAC,GACnB;AA/FnB;AAgGE,QAAM,cAAU,6CAAqB,aAAQ,YAAR,YAAmB,cAAc;AACtE,QAAM,aAAa,UACjB;AAAA,IACE;AAAA,MACE,eAAe,cAAU,kCAAW;AAAA,QAClC,QAAQ,QAAQ;AAAA,QAChB,yBAAyB;AAAA,QACzB,aAAa;AAAA,MACf,CAAC,CAAC;AAAA,MACF,GAAG,QAAQ;AAAA,IACb;AAAA,IACA,kBAAkB,OAAO;AAAA,EAC3B;AASF,QAAM,uBAAuB,CAC3B,WACA,eACuB;AAAA,IACvB,UAAU,WAAW,SAAS;AAAA,IAC9B,KAAK,CAAC,EAAE,KAAK,MAAM;AAEjB,UACE,cAAc,gBACd,uCAAW,SAAS,aACpB,EAAC,uCAAW,SAAS,cACrB;AACA,eAAO,GAAG,SAAS,MAAM,IAAI;AAAA,MAC/B;AACA,aAAO,GAAG,aAAa,OAAO,GAAG,IAAI;AAAA,IACvC;AAAA,IACA,SAAS;AAAA,IACT,OAAO,QAAQ;AAAA,EACjB;AAEA,QAAM,kBAAkB,CAAC,YAAiC;AAExD,UAAM,YAAY,QAAQ;AAE1B,QAAI,WAAW;AAEb,YAAM,qBAAqB,UAAU,SAAS,UAAU;AAExD,UAAI,oBAAoB;AAEtB,eAAO,IAAI,2DAAkC,4BAAW,eAAe;AAAA,UACrE,GAAG,qBAAqB,QAAQ,SAAS;AAAA,UACzC,gBAAgB;AAAA,QAClB,CAAC;AAAA,MACH,WAAW,UAAU,SAAS,UAAU,GAAG;AACzC,cAAM,IAAI;AAAA,UACR;AAAA,QACF;AAAA,MACF;AAAA,IACF;AAGA,WAAO,IAAI,2DAAkC,4BAAW,QAAQ;AAAA,MAC9D,GAAG,qBAAqB,MAAM;AAAA,MAC9B,gBAAgB;AAAA,IAClB,CAAC;AAAA,EACH;AAEA,QAAM,uBAAuB,CAAC,YAAsC;AAElE,UAAM,YAAY,QAAQ;AAC1B,QAAI,CAAC,WAAW;AACd,YAAM,IAAI;AAAA,QACR;AAAA,MACF;AAAA,IACF;AAIA,UAAM,qBAAqB,UAAU,SAAS,OAAO;AAErD,QAAI,oBAAoB;AAEtB,YAAM,QAAQ,IAAI;AAAA,QAChB,4BAAW;AAAA,QACX;AAAA,UACE,GAAG,qBAAqB,aAAa,SAAS;AAAA,UAC9C,gBAAgB;AAAA,QAClB;AAAA,MACF;AAGA,YAAM,uBAAuB,UAAU,QAAQ,YAAY,OAAO;AAGlE,YAAM,oBAAoB,IAAI;AAAA,QAC5B;AAAA,YACA,kCAAW;AAAA,UACT,QAAQ,QAAQ;AAAA,UAChB,yBAAyB;AAAA,UACzB,aAAa;AAAA,QACf,CAAC;AAAA,MACH;AAGA,YAAM,UAAU,OAAM,WAAU;AAC9B,YAAI,CAAC,OAAO,UAAU,CAAC,MAAM,QAAQ,OAAO,MAAM,GAAG;AACnD,gBAAM,IAAI,MAAM,2CAA2C;AAAA,QAC7D;AAGA,cAAM,WAAW,MAAM,kBAAkB;AAAA,UACvC,OAAO;AAAA,UACP,4BAAW;AAAA;AAAA,QACb;AAEA,cAAM,aAAa,SAAS,KAAK,IAAI,CAAC,SAAc,KAAK,SAAS;AAElE,eAAO;AAAA,UACL;AAAA,UACA,OAAO,SAAS,QACZ,EAAE,QAAQ,SAAS,MAAM,aAAa,IACtC;AAAA,UACJ,UAAU,EAAE,SAAS,CAAC,GAAG,MAAM,SAAS;AAAA,UACxC,UAAU,CAAC;AAAA,QACb;AAAA,MACF;AAEA,aAAO;AAAA,IACT,OAAO;AACL,YAAM,IAAI;AAAA,QACR;AAAA,MACF;AAAA,IACF;AAAA,EACF;AAEA,QAAM,WAAW,CAAC,YAAiC,gBAAgB,OAAO;AAE1E,WAAS,uBAAuB;AAChC,WAAS,YAAY;AACrB,WAAS,gBAAgB;AACzB,WAAS,aAAa,CAAC,YAAoB;AACzC,UAAM,IAAI,iCAAiB,EAAE,SAAS,WAAW,aAAa,CAAC;AAAA,EACjE;AACA,WAAS,iBAAiB;AAC1B,WAAS,qBAAqB;AAC9B,SAAO;AACT;AAEO,IAAM,UAAU,cAAc;","names":[]}
+{"version":3,"sources":["../src/baseten-provider.ts","../src/version.ts"],"sourcesContent":["import {\n  OpenAICompatibleChatLanguageModel,\n  OpenAICompatibleEmbeddingModel,\n  ProviderErrorStructure,\n} from '@ai-sdk/openai-compatible';\nimport {\n  EmbeddingModelV4,\n  LanguageModelV4,\n  NoSuchModelError,\n  ProviderV4,\n} from '@ai-sdk/provider';\nimport {\n  FetchFunction,\n  loadApiKey,\n  withoutTrailingSlash,\n  withUserAgentSuffix,\n} from '@ai-sdk/provider-utils';\nimport { z } from 'zod/v4';\nimport { BasetenChatModelId } from './baseten-chat-options';\nimport { BasetenEmbeddingModelId } from './baseten-embedding-options';\nimport { PerformanceClient } from '@basetenlabs/performance-client';\nimport { VERSION } from './version';\n\nexport type BasetenErrorData = z.infer<typeof basetenErrorSchema>;\n\nconst basetenErrorSchema = z.object({\n  error: z.string(),\n});\n\nconst basetenErrorStructure: ProviderErrorStructure<BasetenErrorData> = {\n  errorSchema: basetenErrorSchema,\n  errorToMessage: data => data.error,\n};\n\nexport interface BasetenProviderSettings {\n  /**\n   * Baseten API key. Default value is taken from the `BASETEN_API_KEY`\n   * environment variable.\n   */\n  apiKey?: string;\n\n  /**\n   * Base URL for the Model APIs. Default: 'https://inference.baseten.co/v1'\n   */\n  baseURL?: string;\n\n  /**\n   * Model URL for custom models (chat or embeddings).\n   * If not supplied, the default Model APIs will be used.\n   */\n  modelURL?: string;\n  /**\n   * Custom headers to include in the requests.\n   */\n  headers?: Record<string, string>;\n\n  /**\n   * Custom fetch implementation. You can use it as a middleware to intercept requests,\n   * or to provide a custom fetch implementation for e.g. testing.\n   */\n  fetch?: FetchFunction;\n}\n\nexport interface BasetenProvider extends ProviderV4 {\n  /**\n   * Creates a chat model for text generation.\n   */\n  (modelId?: BasetenChatModelId): LanguageModelV4;\n\n  /**\n   * Creates a chat model for text generation.\n   */\n  chatModel(modelId?: BasetenChatModelId): LanguageModelV4;\n\n  /**\n   * Creates a language model for text generation. Alias for chatModel.\n   */\n  languageModel(modelId?: BasetenChatModelId): LanguageModelV4;\n\n  /**\n   * Creates a embedding model for text generation.\n   */\n  embeddingModel(modelId?: BasetenEmbeddingModelId): EmbeddingModelV4;\n\n  /**\n   * @deprecated Use `embeddingModel` instead.\n   */\n  textEmbeddingModel(modelId?: BasetenEmbeddingModelId): EmbeddingModelV4;\n}\n\n// by default, we use the Model APIs\nconst defaultBaseURL = 'https://inference.baseten.co/v1';\n\nexport function createBaseten(\n  options: BasetenProviderSettings = {},\n): BasetenProvider {\n  const baseURL = withoutTrailingSlash(options.baseURL ?? defaultBaseURL);\n  const getHeaders = () =>\n    withUserAgentSuffix(\n      {\n        Authorization: `Bearer ${loadApiKey({\n          apiKey: options.apiKey,\n          environmentVariableName: 'BASETEN_API_KEY',\n          description: 'Baseten API key',\n        })}`,\n        ...options.headers,\n      },\n      `ai-sdk/baseten/${VERSION}`,\n    );\n\n  interface CommonModelConfig {\n    provider: string;\n    url: ({ path }: { path: string }) => string;\n    headers: () => Record<string, string>;\n    fetch?: FetchFunction;\n  }\n\n  const getCommonModelConfig = (\n    modelType: string,\n    customURL?: string,\n  ): CommonModelConfig => ({\n    provider: `baseten.${modelType}`,\n    url: ({ path }) => {\n      // For embeddings with /sync URLs (but not /sync/v1), we need to add /v1\n      if (\n        modelType === 'embedding' &&\n        customURL?.includes('/sync') &&\n        !customURL?.includes('/sync/v1')\n      ) {\n        return `${customURL}/v1${path}`;\n      }\n      return `${customURL || baseURL}${path}`;\n    },\n    headers: getHeaders,\n    fetch: options.fetch,\n  });\n\n  const createChatModel = (modelId?: BasetenChatModelId) => {\n    // Use modelURL if provided, otherwise use default Model APIs\n    const customURL = options.modelURL;\n\n    if (customURL) {\n      // Check if this is a /sync/v1 endpoint (OpenAI-compatible) or /predict endpoint (custom)\n      const isOpenAICompatible = customURL.includes('/sync/v1');\n\n      if (isOpenAICompatible) {\n        // For /sync/v1 endpoints, use standard OpenAI-compatible format\n        return new OpenAICompatibleChatLanguageModel(modelId ?? 'placeholder', {\n          ...getCommonModelConfig('chat', customURL),\n          errorStructure: basetenErrorStructure,\n        });\n      } else if (customURL.includes('/predict')) {\n        throw new Error(\n          'Not supported. You must use a /sync/v1 endpoint for chat models.',\n        );\n      }\n    }\n\n    // Use default OpenAI-compatible format for Model APIs\n    return new OpenAICompatibleChatLanguageModel(modelId ?? 'chat', {\n      ...getCommonModelConfig('chat'),\n      errorStructure: basetenErrorStructure,\n    });\n  };\n\n  const createEmbeddingModel = (modelId?: BasetenEmbeddingModelId) => {\n    // Use modelURL if provided\n    const customURL = options.modelURL;\n    if (!customURL) {\n      throw new Error(\n        'No model URL provided for embeddings. Please set modelURL option for embeddings.',\n      );\n    }\n\n    // Check if this is a /sync or /sync/v1 endpoint (OpenAI-compatible)\n    // We support both /sync and /sync/v1, stripping /v1 before passing to Performance Client, as Performance Client adds /v1 itself\n    const isOpenAICompatible = customURL.includes('/sync');\n\n    if (isOpenAICompatible) {\n      // Create the model using OpenAICompatibleEmbeddingModel and override doEmbed\n      const model = new OpenAICompatibleEmbeddingModel(\n        modelId ?? 'embeddings',\n        {\n          ...getCommonModelConfig('embedding', customURL),\n          errorStructure: basetenErrorStructure,\n        },\n      );\n\n      // Strip /v1 from URL if present before passing to Performance Client to avoid double /v1\n      const performanceClientURL = customURL.replace('/sync/v1', '/sync');\n\n      // Initialize the B10 Performance Client once for reuse\n      const performanceClient = new PerformanceClient(\n        performanceClientURL,\n        loadApiKey({\n          apiKey: options.apiKey,\n          environmentVariableName: 'BASETEN_API_KEY',\n          description: 'Baseten API key',\n        }),\n      );\n\n      // Override the doEmbed method to use the pre-created Performance Client\n      model.doEmbed = async params => {\n        if (!params.values || !Array.isArray(params.values)) {\n          throw new Error('params.values must be an array of strings');\n        }\n\n        // Performance Client handles batching internally, so we don't need to limit in 128 here\n        const response = await performanceClient.embed(\n          params.values,\n          modelId ?? 'embeddings', // model_id is for Model APIs, we don't use it here for dedicated\n        );\n        // Transform the response to match the expected format\n        const embeddings = response.data.map((item: any) => item.embedding);\n\n        return {\n          embeddings,\n          usage: response.usage\n            ? { tokens: response.usage.total_tokens }\n            : undefined,\n          response: { headers: {}, body: response },\n          warnings: [],\n        };\n      };\n\n      return model;\n    } else {\n      throw new Error(\n        'Not supported. You must use a /sync or /sync/v1 endpoint for embeddings.',\n      );\n    }\n  };\n\n  const provider = (modelId?: BasetenChatModelId) => createChatModel(modelId);\n\n  provider.specificationVersion = 'v4' as const;\n  provider.chatModel = createChatModel;\n  provider.languageModel = createChatModel;\n  provider.imageModel = (modelId: string) => {\n    throw new NoSuchModelError({ modelId, modelType: 'imageModel' });\n  };\n  provider.embeddingModel = createEmbeddingModel;\n  provider.textEmbeddingModel = createEmbeddingModel;\n  return provider;\n}\n\nexport const baseten = createBaseten();\n","// Version string of this package injected at build time.\ndeclare const __PACKAGE_VERSION__: string | undefined;\nexport const VERSION: string =\n  typeof __PACKAGE_VERSION__ !== 'undefined'\n    ? __PACKAGE_VERSION__\n    : '0.0.0-test';\n"],"mappings":";AAAA;AAAA,EACE;AAAA,EACA;AAAA,OAEK;AACP;AAAA,EAGE;AAAA,OAEK;AACP;AAAA,EAEE;AAAA,EACA;AAAA,EACA;AAAA,OACK;AACP,SAAS,SAAS;AAGlB,SAAS,yBAAyB;;;AClB3B,IAAM,UACX,OACI,kBACA;;;ADoBN,IAAM,qBAAqB,EAAE,OAAO;AAAA,EAClC,OAAO,EAAE,OAAO;AAClB,CAAC;AAED,IAAM,wBAAkE;AAAA,EACtE,aAAa;AAAA,EACb,gBAAgB,UAAQ,KAAK;AAC/B;AA2DA,IAAM,iBAAiB;AAEhB,SAAS,cACd,UAAmC,CAAC,GACnB;AA/FnB;AAgGE,QAAM,UAAU,sBAAqB,aAAQ,YAAR,YAAmB,cAAc;AACtE,QAAM,aAAa,MACjB;AAAA,IACE;AAAA,MACE,eAAe,UAAU,WAAW;AAAA,QAClC,QAAQ,QAAQ;AAAA,QAChB,yBAAyB;AAAA,QACzB,aAAa;AAAA,MACf,CAAC,CAAC;AAAA,MACF,GAAG,QAAQ;AAAA,IACb;AAAA,IACA,kBAAkB,OAAO;AAAA,EAC3B;AASF,QAAM,uBAAuB,CAC3B,WACA,eACuB;AAAA,IACvB,UAAU,WAAW,SAAS;AAAA,IAC9B,KAAK,CAAC,EAAE,KAAK,MAAM;AAEjB,UACE,cAAc,gBACd,uCAAW,SAAS,aACpB,EAAC,uCAAW,SAAS,cACrB;AACA,eAAO,GAAG,SAAS,MAAM,IAAI;AAAA,MAC/B;AACA,aAAO,GAAG,aAAa,OAAO,GAAG,IAAI;AAAA,IACvC;AAAA,IACA,SAAS;AAAA,IACT,OAAO,QAAQ;AAAA,EACjB;AAEA,QAAM,kBAAkB,CAAC,YAAiC;AAExD,UAAM,YAAY,QAAQ;AAE1B,QAAI,WAAW;AAEb,YAAM,qBAAqB,UAAU,SAAS,UAAU;AAExD,UAAI,oBAAoB;AAEtB,eAAO,IAAI,kCAAkC,4BAAW,eAAe;AAAA,UACrE,GAAG,qBAAqB,QAAQ,SAAS;AAAA,UACzC,gBAAgB;AAAA,QAClB,CAAC;AAAA,MACH,WAAW,UAAU,SAAS,UAAU,GAAG;AACzC,cAAM,IAAI;AAAA,UACR;AAAA,QACF;AAAA,MACF;AAAA,IACF;AAGA,WAAO,IAAI,kCAAkC,4BAAW,QAAQ;AAAA,MAC9D,GAAG,qBAAqB,MAAM;AAAA,MAC9B,gBAAgB;AAAA,IAClB,CAAC;AAAA,EACH;AAEA,QAAM,uBAAuB,CAAC,YAAsC;AAElE,UAAM,YAAY,QAAQ;AAC1B,QAAI,CAAC,WAAW;AACd,YAAM,IAAI;AAAA,QACR;AAAA,MACF;AAAA,IACF;AAIA,UAAM,qBAAqB,UAAU,SAAS,OAAO;AAErD,QAAI,oBAAoB;AAEtB,YAAM,QAAQ,IAAI;AAAA,QAChB,4BAAW;AAAA,QACX;AAAA,UACE,GAAG,qBAAqB,aAAa,SAAS;AAAA,UAC9C,gBAAgB;AAAA,QAClB;AAAA,MACF;AAGA,YAAM,uBAAuB,UAAU,QAAQ,YAAY,OAAO;AAGlE,YAAM,oBAAoB,IAAI;AAAA,QAC5B;AAAA,QACA,WAAW;AAAA,UACT,QAAQ,QAAQ;AAAA,UAChB,yBAAyB;AAAA,UACzB,aAAa;AAAA,QACf,CAAC;AAAA,MACH;AAGA,YAAM,UAAU,OAAM,WAAU;AAC9B,YAAI,CAAC,OAAO,UAAU,CAAC,MAAM,QAAQ,OAAO,MAAM,GAAG;AACnD,gBAAM,IAAI,MAAM,2CAA2C;AAAA,QAC7D;AAGA,cAAM,WAAW,MAAM,kBAAkB;AAAA,UACvC,OAAO;AAAA,UACP,4BAAW;AAAA;AAAA,QACb;AAEA,cAAM,aAAa,SAAS,KAAK,IAAI,CAAC,SAAc,KAAK,SAAS;AAElE,eAAO;AAAA,UACL;AAAA,UACA,OAAO,SAAS,QACZ,EAAE,QAAQ,SAAS,MAAM,aAAa,IACtC;AAAA,UACJ,UAAU,EAAE,SAAS,CAAC,GAAG,MAAM,SAAS;AAAA,UACxC,UAAU,CAAC;AAAA,QACb;AAAA,MACF;AAEA,aAAO;AAAA,IACT,OAAO;AACL,YAAM,IAAI;AAAA,QACR;AAAA,MACF;AAAA,IACF;AAAA,EACF;AAEA,QAAM,WAAW,CAAC,YAAiC,gBAAgB,OAAO;AAE1E,WAAS,uBAAuB;AAChC,WAAS,YAAY;AACrB,WAAS,gBAAgB;AACzB,WAAS,aAAa,CAAC,YAAoB;AACzC,UAAM,IAAI,iBAAiB,EAAE,SAAS,WAAW,aAAa,CAAC;AAAA,EACjE;AACA,WAAS,iBAAiB;AAC1B,WAAS,qBAAqB;AAC9B,SAAO;AACT;AAEO,IAAM,UAAU,cAAc;","names":[]}

package/package.json

@@ -1,10 +1,10 @@
 {
   "name": "@ai-sdk/baseten",
-  "version": "2.0.0-beta.23",
+  "version": "2.0.0-beta.25",
+  "type": "module",
   "license": "Apache-2.0",
   "sideEffects": false,
   "main": "./dist/index.js",
-  "module": "./dist/index.mjs",
   "types": "./dist/index.d.ts",
   "files": [
     "dist/**/*",
@@ -24,15 +24,15 @@
     "./package.json": "./package.json",
     ".": {
       "types": "./dist/index.d.ts",
-      "import": "./dist/index.mjs",
-      "require": "./dist/index.js"
+      "import": "./dist/index.js",
+      "default": "./dist/index.js"
     }
   },
   "dependencies": {
     "@basetenlabs/performance-client": "^0.0.10",
-    "@ai-sdk/openai-compatible": "3.0.0-beta.23",
-    "@ai-sdk/provider": "4.0.0-beta.10",
-    "@ai-sdk/provider-utils": "5.0.0-beta.18"
+    "@ai-sdk/openai-compatible": "3.0.0-beta.25",
+    "@ai-sdk/provider": "4.0.0-beta.12",
+    "@ai-sdk/provider-utils": "5.0.0-beta.20"
   },
   "devDependencies": {
     "@types/node": "20.17.24",

package/dist/index.d.mts

@@ -1,67 +0,0 @@
-import { ProviderV4, LanguageModelV4, EmbeddingModelV4 } from '@ai-sdk/provider';
-import { FetchFunction } from '@ai-sdk/provider-utils';
-import { z } from 'zod/v4';
-
-type BasetenChatModelId = 'deepseek-ai/DeepSeek-R1-0528' | 'deepseek-ai/DeepSeek-V3-0324' | 'deepseek-ai/DeepSeek-V3.1' | 'moonshotai/Kimi-K2-Instruct-0905' | 'moonshotai/Kimi-K2-Thinking' | 'Qwen/Qwen3-235B-A22B-Instruct-2507' | 'Qwen/Qwen3-Coder-480B-A35B-Instruct' | 'openai/gpt-oss-120b' | 'zai-org/GLM-4.6' | 'zai-org/GLM-4.7' | (string & {});
-
-type BasetenEmbeddingModelId = string & {};
-declare const basetenEmbeddingModelOptions: z.ZodObject<{}, z.core.$strip>;
-type BasetenEmbeddingModelOptions = z.infer<typeof basetenEmbeddingModelOptions>;
-
-type BasetenErrorData = z.infer<typeof basetenErrorSchema>;
-declare const basetenErrorSchema: z.ZodObject<{
-    error: z.ZodString;
-}, z.core.$strip>;
-interface BasetenProviderSettings {
-    /**
-     * Baseten API key. Default value is taken from the `BASETEN_API_KEY`
-     * environment variable.
-     */
-    apiKey?: string;
-    /**
-     * Base URL for the Model APIs. Default: 'https://inference.baseten.co/v1'
-     */
-    baseURL?: string;
-    /**
-     * Model URL for custom models (chat or embeddings).
-     * If not supplied, the default Model APIs will be used.
-     */
-    modelURL?: string;
-    /**
-     * Custom headers to include in the requests.
-     */
-    headers?: Record<string, string>;
-    /**
-     * Custom fetch implementation. You can use it as a middleware to intercept requests,
-     * or to provide a custom fetch implementation for e.g. testing.
-     */
-    fetch?: FetchFunction;
-}
-interface BasetenProvider extends ProviderV4 {
-    /**
-     * Creates a chat model for text generation.
-     */
-    (modelId?: BasetenChatModelId): LanguageModelV4;
-    /**
-     * Creates a chat model for text generation.
-     */
-    chatModel(modelId?: BasetenChatModelId): LanguageModelV4;
-    /**
-     * Creates a language model for text generation. Alias for chatModel.
-     */
-    languageModel(modelId?: BasetenChatModelId): LanguageModelV4;
-    /**
-     * Creates a embedding model for text generation.
-     */
-    embeddingModel(modelId?: BasetenEmbeddingModelId): EmbeddingModelV4;
-    /**
-     * @deprecated Use `embeddingModel` instead.
-     */
-    textEmbeddingModel(modelId?: BasetenEmbeddingModelId): EmbeddingModelV4;
-}
-declare function createBaseten(options?: BasetenProviderSettings): BasetenProvider;
-declare const baseten: BasetenProvider;
-
-declare const VERSION: string;
-
-export { type BasetenChatModelId, type BasetenEmbeddingModelOptions, type BasetenErrorData, type BasetenProvider, type BasetenProviderSettings, VERSION, baseten, createBaseten };

package/dist/index.mjs

@@ -1,140 +0,0 @@
-// src/baseten-provider.ts
-import {
-  OpenAICompatibleChatLanguageModel,
-  OpenAICompatibleEmbeddingModel
-} from "@ai-sdk/openai-compatible";
-import {
-  NoSuchModelError
-} from "@ai-sdk/provider";
-import {
-  loadApiKey,
-  withoutTrailingSlash,
-  withUserAgentSuffix
-} from "@ai-sdk/provider-utils";
-import { z } from "zod/v4";
-import { PerformanceClient } from "@basetenlabs/performance-client";
-
-// src/version.ts
-var VERSION = true ? "2.0.0-beta.23" : "0.0.0-test";
-
-// src/baseten-provider.ts
-var basetenErrorSchema = z.object({
-  error: z.string()
-});
-var basetenErrorStructure = {
-  errorSchema: basetenErrorSchema,
-  errorToMessage: (data) => data.error
-};
-var defaultBaseURL = "https://inference.baseten.co/v1";
-function createBaseten(options = {}) {
-  var _a;
-  const baseURL = withoutTrailingSlash((_a = options.baseURL) != null ? _a : defaultBaseURL);
-  const getHeaders = () => withUserAgentSuffix(
-    {
-      Authorization: `Bearer ${loadApiKey({
-        apiKey: options.apiKey,
-        environmentVariableName: "BASETEN_API_KEY",
-        description: "Baseten API key"
-      })}`,
-      ...options.headers
-    },
-    `ai-sdk/baseten/${VERSION}`
-  );
-  const getCommonModelConfig = (modelType, customURL) => ({
-    provider: `baseten.${modelType}`,
-    url: ({ path }) => {
-      if (modelType === "embedding" && (customURL == null ? void 0 : customURL.includes("/sync")) && !(customURL == null ? void 0 : customURL.includes("/sync/v1"))) {
-        return `${customURL}/v1${path}`;
-      }
-      return `${customURL || baseURL}${path}`;
-    },
-    headers: getHeaders,
-    fetch: options.fetch
-  });
-  const createChatModel = (modelId) => {
-    const customURL = options.modelURL;
-    if (customURL) {
-      const isOpenAICompatible = customURL.includes("/sync/v1");
-      if (isOpenAICompatible) {
-        return new OpenAICompatibleChatLanguageModel(modelId != null ? modelId : "placeholder", {
-          ...getCommonModelConfig("chat", customURL),
-          errorStructure: basetenErrorStructure
-        });
-      } else if (customURL.includes("/predict")) {
-        throw new Error(
-          "Not supported. You must use a /sync/v1 endpoint for chat models."
-        );
-      }
-    }
-    return new OpenAICompatibleChatLanguageModel(modelId != null ? modelId : "chat", {
-      ...getCommonModelConfig("chat"),
-      errorStructure: basetenErrorStructure
-    });
-  };
-  const createEmbeddingModel = (modelId) => {
-    const customURL = options.modelURL;
-    if (!customURL) {
-      throw new Error(
-        "No model URL provided for embeddings. Please set modelURL option for embeddings."
-      );
-    }
-    const isOpenAICompatible = customURL.includes("/sync");
-    if (isOpenAICompatible) {
-      const model = new OpenAICompatibleEmbeddingModel(
-        modelId != null ? modelId : "embeddings",
-        {
-          ...getCommonModelConfig("embedding", customURL),
-          errorStructure: basetenErrorStructure
-        }
-      );
-      const performanceClientURL = customURL.replace("/sync/v1", "/sync");
-      const performanceClient = new PerformanceClient(
-        performanceClientURL,
-        loadApiKey({
-          apiKey: options.apiKey,
-          environmentVariableName: "BASETEN_API_KEY",
-          description: "Baseten API key"
-        })
-      );
-      model.doEmbed = async (params) => {
-        if (!params.values || !Array.isArray(params.values)) {
-          throw new Error("params.values must be an array of strings");
-        }
-        const response = await performanceClient.embed(
-          params.values,
-          modelId != null ? modelId : "embeddings"
-          // model_id is for Model APIs, we don't use it here for dedicated
-        );
-        const embeddings = response.data.map((item) => item.embedding);
-        return {
-          embeddings,
-          usage: response.usage ? { tokens: response.usage.total_tokens } : void 0,
-          response: { headers: {}, body: response },
-          warnings: []
-        };
-      };
-      return model;
-    } else {
-      throw new Error(
-        "Not supported. You must use a /sync or /sync/v1 endpoint for embeddings."
-      );
-    }
-  };
-  const provider = (modelId) => createChatModel(modelId);
-  provider.specificationVersion = "v4";
-  provider.chatModel = createChatModel;
-  provider.languageModel = createChatModel;
-  provider.imageModel = (modelId) => {
-    throw new NoSuchModelError({ modelId, modelType: "imageModel" });
-  };
-  provider.embeddingModel = createEmbeddingModel;
-  provider.textEmbeddingModel = createEmbeddingModel;
-  return provider;
-}
-var baseten = createBaseten();
-export {
-  VERSION,
-  baseten,
-  createBaseten
-};
-//# sourceMappingURL=index.mjs.map

package/dist/index.mjs.map

@@ -1 +0,0 @@
-{"version":3,"sources":["../src/baseten-provider.ts","../src/version.ts"],"sourcesContent":["import {\n  OpenAICompatibleChatLanguageModel,\n  OpenAICompatibleEmbeddingModel,\n  ProviderErrorStructure,\n} from '@ai-sdk/openai-compatible';\nimport {\n  EmbeddingModelV4,\n  LanguageModelV4,\n  NoSuchModelError,\n  ProviderV4,\n} from '@ai-sdk/provider';\nimport {\n  FetchFunction,\n  loadApiKey,\n  withoutTrailingSlash,\n  withUserAgentSuffix,\n} from '@ai-sdk/provider-utils';\nimport { z } from 'zod/v4';\nimport { BasetenChatModelId } from './baseten-chat-options';\nimport { BasetenEmbeddingModelId } from './baseten-embedding-options';\nimport { PerformanceClient } from '@basetenlabs/performance-client';\nimport { VERSION } from './version';\n\nexport type BasetenErrorData = z.infer<typeof basetenErrorSchema>;\n\nconst basetenErrorSchema = z.object({\n  error: z.string(),\n});\n\nconst basetenErrorStructure: ProviderErrorStructure<BasetenErrorData> = {\n  errorSchema: basetenErrorSchema,\n  errorToMessage: data => data.error,\n};\n\nexport interface BasetenProviderSettings {\n  /**\n   * Baseten API key. Default value is taken from the `BASETEN_API_KEY`\n   * environment variable.\n   */\n  apiKey?: string;\n\n  /**\n   * Base URL for the Model APIs. Default: 'https://inference.baseten.co/v1'\n   */\n  baseURL?: string;\n\n  /**\n   * Model URL for custom models (chat or embeddings).\n   * If not supplied, the default Model APIs will be used.\n   */\n  modelURL?: string;\n  /**\n   * Custom headers to include in the requests.\n   */\n  headers?: Record<string, string>;\n\n  /**\n   * Custom fetch implementation. You can use it as a middleware to intercept requests,\n   * or to provide a custom fetch implementation for e.g. testing.\n   */\n  fetch?: FetchFunction;\n}\n\nexport interface BasetenProvider extends ProviderV4 {\n  /**\n   * Creates a chat model for text generation.\n   */\n  (modelId?: BasetenChatModelId): LanguageModelV4;\n\n  /**\n   * Creates a chat model for text generation.\n   */\n  chatModel(modelId?: BasetenChatModelId): LanguageModelV4;\n\n  /**\n   * Creates a language model for text generation. Alias for chatModel.\n   */\n  languageModel(modelId?: BasetenChatModelId): LanguageModelV4;\n\n  /**\n   * Creates a embedding model for text generation.\n   */\n  embeddingModel(modelId?: BasetenEmbeddingModelId): EmbeddingModelV4;\n\n  /**\n   * @deprecated Use `embeddingModel` instead.\n   */\n  textEmbeddingModel(modelId?: BasetenEmbeddingModelId): EmbeddingModelV4;\n}\n\n// by default, we use the Model APIs\nconst defaultBaseURL = 'https://inference.baseten.co/v1';\n\nexport function createBaseten(\n  options: BasetenProviderSettings = {},\n): BasetenProvider {\n  const baseURL = withoutTrailingSlash(options.baseURL ?? defaultBaseURL);\n  const getHeaders = () =>\n    withUserAgentSuffix(\n      {\n        Authorization: `Bearer ${loadApiKey({\n          apiKey: options.apiKey,\n          environmentVariableName: 'BASETEN_API_KEY',\n          description: 'Baseten API key',\n        })}`,\n        ...options.headers,\n      },\n      `ai-sdk/baseten/${VERSION}`,\n    );\n\n  interface CommonModelConfig {\n    provider: string;\n    url: ({ path }: { path: string }) => string;\n    headers: () => Record<string, string>;\n    fetch?: FetchFunction;\n  }\n\n  const getCommonModelConfig = (\n    modelType: string,\n    customURL?: string,\n  ): CommonModelConfig => ({\n    provider: `baseten.${modelType}`,\n    url: ({ path }) => {\n      // For embeddings with /sync URLs (but not /sync/v1), we need to add /v1\n      if (\n        modelType === 'embedding' &&\n        customURL?.includes('/sync') &&\n        !customURL?.includes('/sync/v1')\n      ) {\n        return `${customURL}/v1${path}`;\n      }\n      return `${customURL || baseURL}${path}`;\n    },\n    headers: getHeaders,\n    fetch: options.fetch,\n  });\n\n  const createChatModel = (modelId?: BasetenChatModelId) => {\n    // Use modelURL if provided, otherwise use default Model APIs\n    const customURL = options.modelURL;\n\n    if (customURL) {\n      // Check if this is a /sync/v1 endpoint (OpenAI-compatible) or /predict endpoint (custom)\n      const isOpenAICompatible = customURL.includes('/sync/v1');\n\n      if (isOpenAICompatible) {\n        // For /sync/v1 endpoints, use standard OpenAI-compatible format\n        return new OpenAICompatibleChatLanguageModel(modelId ?? 'placeholder', {\n          ...getCommonModelConfig('chat', customURL),\n          errorStructure: basetenErrorStructure,\n        });\n      } else if (customURL.includes('/predict')) {\n        throw new Error(\n          'Not supported. You must use a /sync/v1 endpoint for chat models.',\n        );\n      }\n    }\n\n    // Use default OpenAI-compatible format for Model APIs\n    return new OpenAICompatibleChatLanguageModel(modelId ?? 'chat', {\n      ...getCommonModelConfig('chat'),\n      errorStructure: basetenErrorStructure,\n    });\n  };\n\n  const createEmbeddingModel = (modelId?: BasetenEmbeddingModelId) => {\n    // Use modelURL if provided\n    const customURL = options.modelURL;\n    if (!customURL) {\n      throw new Error(\n        'No model URL provided for embeddings. Please set modelURL option for embeddings.',\n      );\n    }\n\n    // Check if this is a /sync or /sync/v1 endpoint (OpenAI-compatible)\n    // We support both /sync and /sync/v1, stripping /v1 before passing to Performance Client, as Performance Client adds /v1 itself\n    const isOpenAICompatible = customURL.includes('/sync');\n\n    if (isOpenAICompatible) {\n      // Create the model using OpenAICompatibleEmbeddingModel and override doEmbed\n      const model = new OpenAICompatibleEmbeddingModel(\n        modelId ?? 'embeddings',\n        {\n          ...getCommonModelConfig('embedding', customURL),\n          errorStructure: basetenErrorStructure,\n        },\n      );\n\n      // Strip /v1 from URL if present before passing to Performance Client to avoid double /v1\n      const performanceClientURL = customURL.replace('/sync/v1', '/sync');\n\n      // Initialize the B10 Performance Client once for reuse\n      const performanceClient = new PerformanceClient(\n        performanceClientURL,\n        loadApiKey({\n          apiKey: options.apiKey,\n          environmentVariableName: 'BASETEN_API_KEY',\n          description: 'Baseten API key',\n        }),\n      );\n\n      // Override the doEmbed method to use the pre-created Performance Client\n      model.doEmbed = async params => {\n        if (!params.values || !Array.isArray(params.values)) {\n          throw new Error('params.values must be an array of strings');\n        }\n\n        // Performance Client handles batching internally, so we don't need to limit in 128 here\n        const response = await performanceClient.embed(\n          params.values,\n          modelId ?? 'embeddings', // model_id is for Model APIs, we don't use it here for dedicated\n        );\n        // Transform the response to match the expected format\n        const embeddings = response.data.map((item: any) => item.embedding);\n\n        return {\n          embeddings,\n          usage: response.usage\n            ? { tokens: response.usage.total_tokens }\n            : undefined,\n          response: { headers: {}, body: response },\n          warnings: [],\n        };\n      };\n\n      return model;\n    } else {\n      throw new Error(\n        'Not supported. You must use a /sync or /sync/v1 endpoint for embeddings.',\n      );\n    }\n  };\n\n  const provider = (modelId?: BasetenChatModelId) => createChatModel(modelId);\n\n  provider.specificationVersion = 'v4' as const;\n  provider.chatModel = createChatModel;\n  provider.languageModel = createChatModel;\n  provider.imageModel = (modelId: string) => {\n    throw new NoSuchModelError({ modelId, modelType: 'imageModel' });\n  };\n  provider.embeddingModel = createEmbeddingModel;\n  provider.textEmbeddingModel = createEmbeddingModel;\n  return provider;\n}\n\nexport const baseten = createBaseten();\n","// Version string of this package injected at build time.\ndeclare const __PACKAGE_VERSION__: string | undefined;\nexport const VERSION: string =\n  typeof __PACKAGE_VERSION__ !== 'undefined'\n    ? __PACKAGE_VERSION__\n    : '0.0.0-test';\n"],"mappings":";AAAA;AAAA,EACE;AAAA,EACA;AAAA,OAEK;AACP;AAAA,EAGE;AAAA,OAEK;AACP;AAAA,EAEE;AAAA,EACA;AAAA,EACA;AAAA,OACK;AACP,SAAS,SAAS;AAGlB,SAAS,yBAAyB;;;AClB3B,IAAM,UACX,OACI,kBACA;;;ADoBN,IAAM,qBAAqB,EAAE,OAAO;AAAA,EAClC,OAAO,EAAE,OAAO;AAClB,CAAC;AAED,IAAM,wBAAkE;AAAA,EACtE,aAAa;AAAA,EACb,gBAAgB,UAAQ,KAAK;AAC/B;AA2DA,IAAM,iBAAiB;AAEhB,SAAS,cACd,UAAmC,CAAC,GACnB;AA/FnB;AAgGE,QAAM,UAAU,sBAAqB,aAAQ,YAAR,YAAmB,cAAc;AACtE,QAAM,aAAa,MACjB;AAAA,IACE;AAAA,MACE,eAAe,UAAU,WAAW;AAAA,QAClC,QAAQ,QAAQ;AAAA,QAChB,yBAAyB;AAAA,QACzB,aAAa;AAAA,MACf,CAAC,CAAC;AAAA,MACF,GAAG,QAAQ;AAAA,IACb;AAAA,IACA,kBAAkB,OAAO;AAAA,EAC3B;AASF,QAAM,uBAAuB,CAC3B,WACA,eACuB;AAAA,IACvB,UAAU,WAAW,SAAS;AAAA,IAC9B,KAAK,CAAC,EAAE,KAAK,MAAM;AAEjB,UACE,cAAc,gBACd,uCAAW,SAAS,aACpB,EAAC,uCAAW,SAAS,cACrB;AACA,eAAO,GAAG,SAAS,MAAM,IAAI;AAAA,MAC/B;AACA,aAAO,GAAG,aAAa,OAAO,GAAG,IAAI;AAAA,IACvC;AAAA,IACA,SAAS;AAAA,IACT,OAAO,QAAQ;AAAA,EACjB;AAEA,QAAM,kBAAkB,CAAC,YAAiC;AAExD,UAAM,YAAY,QAAQ;AAE1B,QAAI,WAAW;AAEb,YAAM,qBAAqB,UAAU,SAAS,UAAU;AAExD,UAAI,oBAAoB;AAEtB,eAAO,IAAI,kCAAkC,4BAAW,eAAe;AAAA,UACrE,GAAG,qBAAqB,QAAQ,SAAS;AAAA,UACzC,gBAAgB;AAAA,QAClB,CAAC;AAAA,MACH,WAAW,UAAU,SAAS,UAAU,GAAG;AACzC,cAAM,IAAI;AAAA,UACR;AAAA,QACF;AAAA,MACF;AAAA,IACF;AAGA,WAAO,IAAI,kCAAkC,4BAAW,QAAQ;AAAA,MAC9D,GAAG,qBAAqB,MAAM;AAAA,MAC9B,gBAAgB;AAAA,IAClB,CAAC;AAAA,EACH;AAEA,QAAM,uBAAuB,CAAC,YAAsC;AAElE,UAAM,YAAY,QAAQ;AAC1B,QAAI,CAAC,WAAW;AACd,YAAM,IAAI;AAAA,QACR;AAAA,MACF;AAAA,IACF;AAIA,UAAM,qBAAqB,UAAU,SAAS,OAAO;AAErD,QAAI,oBAAoB;AAEtB,YAAM,QAAQ,IAAI;AAAA,QAChB,4BAAW;AAAA,QACX;AAAA,UACE,GAAG,qBAAqB,aAAa,SAAS;AAAA,UAC9C,gBAAgB;AAAA,QAClB;AAAA,MACF;AAGA,YAAM,uBAAuB,UAAU,QAAQ,YAAY,OAAO;AAGlE,YAAM,oBAAoB,IAAI;AAAA,QAC5B;AAAA,QACA,WAAW;AAAA,UACT,QAAQ,QAAQ;AAAA,UAChB,yBAAyB;AAAA,UACzB,aAAa;AAAA,QACf,CAAC;AAAA,MACH;AAGA,YAAM,UAAU,OAAM,WAAU;AAC9B,YAAI,CAAC,OAAO,UAAU,CAAC,MAAM,QAAQ,OAAO,MAAM,GAAG;AACnD,gBAAM,IAAI,MAAM,2CAA2C;AAAA,QAC7D;AAGA,cAAM,WAAW,MAAM,kBAAkB;AAAA,UACvC,OAAO;AAAA,UACP,4BAAW;AAAA;AAAA,QACb;AAEA,cAAM,aAAa,SAAS,KAAK,IAAI,CAAC,SAAc,KAAK,SAAS;AAElE,eAAO;AAAA,UACL;AAAA,UACA,OAAO,SAAS,QACZ,EAAE,QAAQ,SAAS,MAAM,aAAa,IACtC;AAAA,UACJ,UAAU,EAAE,SAAS,CAAC,GAAG,MAAM,SAAS;AAAA,UACxC,UAAU,CAAC;AAAA,QACb;AAAA,MACF;AAEA,aAAO;AAAA,IACT,OAAO;AACL,YAAM,IAAI;AAAA,QACR;AAAA,MACF;AAAA,IACF;AAAA,EACF;AAEA,QAAM,WAAW,CAAC,YAAiC,gBAAgB,OAAO;AAE1E,WAAS,uBAAuB;AAChC,WAAS,YAAY;AACrB,WAAS,gBAAgB;AACzB,WAAS,aAAa,CAAC,YAAoB;AACzC,UAAM,IAAI,iBAAiB,EAAE,SAAS,WAAW,aAAa,CAAC;AAAA,EACjE;AACA,WAAS,iBAAiB;AAC1B,WAAS,qBAAqB;AAC9B,SAAO;AACT;AAEO,IAAM,UAAU,cAAc;","names":[]}