@ai-sdk/openai 0.0.67 → 0.0.70

package/CHANGELOG.md

@@ -1,5 +1,33 @@
 # @ai-sdk/openai
 
+## 0.0.70
+
+### Patch Changes
+
+- 3b1b69a: feat: provider-defined tools
+- Updated dependencies [aa98cdb]
+- Updated dependencies [1486128]
+- Updated dependencies [7b937c5]
+- Updated dependencies [3b1b69a]
+- Updated dependencies [811a317]
+  - @ai-sdk/provider-utils@1.0.22
+  - @ai-sdk/provider@0.0.26
+
+## 0.0.69
+
+### Patch Changes
+
+- b9b0d7b: feat (ai): access raw request body
+- Updated dependencies [b9b0d7b]
+  - @ai-sdk/provider@0.0.25
+  - @ai-sdk/provider-utils@1.0.21
+
+## 0.0.68
+
+### Patch Changes
+
+- 741ca51: feat (provider/openai): support mp3 and wav audio inputs
+
 ## 0.0.67
 
 ### Patch Changes

package/dist/index.d.mts

@@ -1,7 +1,7 @@
 import { LanguageModelV1, ProviderV1, EmbeddingModelV1 } from '@ai-sdk/provider';
 import { FetchFunction } from '@ai-sdk/provider-utils';
 
-type OpenAIChatModelId = 'o1-preview' | 'o1-mini' | 'gpt-4o' | 'gpt-4o-2024-05-13' | 'gpt-4o-2024-08-06' | '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-3.5-turbo-0125' | 'gpt-3.5-turbo' | 'gpt-3.5-turbo-1106' | (string & {});
+type OpenAIChatModelId = 'o1-preview' | 'o1-mini' | 'gpt-4o' | 'gpt-4o-2024-05-13' | 'gpt-4o-2024-08-06' | 'gpt-4o-audio-preview' | 'gpt-4o-audio-preview-2024-10-01' | '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-3.5-turbo-0125' | 'gpt-3.5-turbo' | 'gpt-3.5-turbo-1106' | (string & {});
 interface OpenAIChatSettings {
     /**
   Modify the likelihood of specified tokens appearing in the completion.

package/dist/index.d.ts

@@ -1,7 +1,7 @@
 import { LanguageModelV1, ProviderV1, EmbeddingModelV1 } from '@ai-sdk/provider';
 import { FetchFunction } from '@ai-sdk/provider-utils';
 
-type OpenAIChatModelId = 'o1-preview' | 'o1-mini' | 'gpt-4o' | 'gpt-4o-2024-05-13' | 'gpt-4o-2024-08-06' | '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-3.5-turbo-0125' | 'gpt-3.5-turbo' | 'gpt-3.5-turbo-1106' | (string & {});
+type OpenAIChatModelId = 'o1-preview' | 'o1-mini' | 'gpt-4o' | 'gpt-4o-2024-05-13' | 'gpt-4o-2024-08-06' | 'gpt-4o-audio-preview' | 'gpt-4o-audio-preview-2024-10-01' | '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-3.5-turbo-0125' | 'gpt-3.5-turbo' | 'gpt-3.5-turbo-1106' | (string & {});
 interface OpenAIChatSettings {
     /**
   Modify the likelihood of specified tokens appearing in the completion.

package/dist/index.js

@@ -30,7 +30,7 @@ module.exports = __toCommonJS(src_exports);
 var import_provider_utils5 = require("@ai-sdk/provider-utils");
 
 // src/openai-chat-language-model.ts
-var import_provider2 = require("@ai-sdk/provider");
+var import_provider3 = require("@ai-sdk/provider");
 var import_provider_utils3 = require("@ai-sdk/provider-utils");
 var import_zod2 = require("zod");
 
@@ -72,9 +72,31 @@ function convertToOpenAIChatMessages({
                 };
               }
               case "file": {
-                throw new import_provider.UnsupportedFunctionalityError({
-                  functionality: "File content parts in user messages"
-                });
+                if (part.data instanceof URL) {
+                  throw new import_provider.UnsupportedFunctionalityError({
+                    functionality: "'File content parts with URL data' functionality not supported."
+                  });
+                }
+                switch (part.mimeType) {
+                  case "audio/wav": {
+                    return {
+                      type: "input_audio",
+                      input_audio: { data: part.data, format: "wav" }
+                    };
+                  }
+                  case "audio/mp3":
+                  case "audio/mpeg": {
+                    return {
+                      type: "input_audio",
+                      input_audio: { data: part.data, format: "mp3" }
+                    };
+                  }
+                  default: {
+                    throw new import_provider.UnsupportedFunctionalityError({
+                      functionality: `File content part type ${part.mimeType} in user messages`
+                    });
+                  }
+                }
               }
             }
           })
@@ -216,6 +238,105 @@ function getResponseMetadata({
   };
 }
 
+// src/openai-prepare-tools.ts
+var import_provider2 = require("@ai-sdk/provider");
+function prepareTools({
+  mode,
+  useLegacyFunctionCalling = false,
+  structuredOutputs = false
+}) {
+  var _a;
+  const tools = ((_a = mode.tools) == null ? void 0 : _a.length) ? mode.tools : void 0;
+  const toolWarnings = [];
+  if (tools == null) {
+    return { tools: void 0, tool_choice: void 0, toolWarnings };
+  }
+  const toolChoice = mode.toolChoice;
+  if (useLegacyFunctionCalling) {
+    const openaiFunctions = [];
+    for (const tool of tools) {
+      if (tool.type === "provider-defined") {
+        toolWarnings.push({ type: "unsupported-tool", tool });
+      } else {
+        openaiFunctions.push({
+          name: tool.name,
+          description: tool.description,
+          parameters: tool.parameters
+        });
+      }
+    }
+    if (toolChoice == null) {
+      return {
+        functions: openaiFunctions,
+        function_call: void 0,
+        toolWarnings
+      };
+    }
+    const type2 = toolChoice.type;
+    switch (type2) {
+      case "auto":
+      case "none":
+      case void 0:
+        return {
+          functions: openaiFunctions,
+          function_call: void 0,
+          toolWarnings
+        };
+      case "required":
+        throw new import_provider2.UnsupportedFunctionalityError({
+          functionality: "useLegacyFunctionCalling and toolChoice: required"
+        });
+      default:
+        return {
+          functions: openaiFunctions,
+          function_call: { name: toolChoice.toolName },
+          toolWarnings
+        };
+    }
+  }
+  const openaiTools = [];
+  for (const tool of tools) {
+    if (tool.type === "provider-defined") {
+      toolWarnings.push({ type: "unsupported-tool", tool });
+    } else {
+      openaiTools.push({
+        type: "function",
+        function: {
+          name: tool.name,
+          description: tool.description,
+          parameters: tool.parameters,
+          strict: structuredOutputs === true ? true : void 0
+        }
+      });
+    }
+  }
+  if (toolChoice == null) {
+    return { tools: openaiTools, tool_choice: void 0, toolWarnings };
+  }
+  const type = toolChoice.type;
+  switch (type) {
+    case "auto":
+    case "none":
+    case "required":
+      return { tools: openaiTools, tool_choice: type, toolWarnings };
+    case "tool":
+      return {
+        tools: openaiTools,
+        tool_choice: {
+          type: "function",
+          function: {
+            name: toolChoice.toolName
+          }
+        },
+        toolWarnings
+      };
+    default: {
+      const _exhaustiveCheck = type;
+      throw new Error(`Unsupported tool choice type: ${_exhaustiveCheck}`);
+    }
+  }
+}
+
 // src/openai-chat-language-model.ts
 var OpenAIChatLanguageModel = class {
   constructor(modelId, settings, config) {
@@ -228,6 +349,9 @@ var OpenAIChatLanguageModel = class {
     return this.settings.structuredOutputs === true;
   }
   get defaultObjectGenerationMode() {
+    if (isAudioModel(this.modelId)) {
+      return "tool";
+    }
     return this.supportsStructuredOutputs ? "json" : "tool";
   }
   get provider() {
@@ -268,12 +392,12 @@ var OpenAIChatLanguageModel = class {
     }
     const useLegacyFunctionCalling = this.settings.useLegacyFunctionCalling;
     if (useLegacyFunctionCalling && this.settings.parallelToolCalls === true) {
-      throw new import_provider2.UnsupportedFunctionalityError({
+      throw new import_provider3.UnsupportedFunctionalityError({
         functionality: "useLegacyFunctionCalling with parallelToolCalls"
       });
     }
     if (useLegacyFunctionCalling && this.settings.structuredOutputs === true) {
-      throw new import_provider2.UnsupportedFunctionalityError({
+      throw new import_provider3.UnsupportedFunctionalityError({
         functionality: "structuredOutputs with useLegacyFunctionCalling"
       });
     }
@@ -314,16 +438,20 @@ var OpenAIChatLanguageModel = class {
     }
     switch (type) {
       case "regular": {
+        const { tools, tool_choice, functions, function_call, toolWarnings } = prepareTools({
+          mode,
+          useLegacyFunctionCalling,
+          structuredOutputs: this.settings.structuredOutputs
+        });
         return {
           args: {
             ...baseArgs,
-            ...prepareToolsAndToolChoice({
-              mode,
-              useLegacyFunctionCalling,
-              structuredOutputs: this.settings.structuredOutputs
-            })
+            tools,
+            tool_choice,
+            functions,
+            function_call
           },
-          warnings
+          warnings: [...warnings, ...toolWarnings]
         };
       }
       case "object-json": {
@@ -386,14 +514,14 @@ var OpenAIChatLanguageModel = class {
   }
   async doGenerate(options) {
     var _a, _b, _c, _d, _e, _f, _g, _h, _i, _j, _k, _l, _m, _n, _o, _p, _q, _r;
-    const { args, warnings } = this.getArgs(options);
+    const { args: body, warnings } = this.getArgs(options);
     const { responseHeaders, value: response } = await (0, import_provider_utils3.postJsonToApi)({
       url: this.config.url({
         path: "/chat/completions",
         modelId: this.modelId
       }),
       headers: (0, import_provider_utils3.combineHeaders)(this.config.headers(), options.headers),
-      body: args,
+      body,
       failedResponseHandler: openaiFailedResponseHandler,
       successfulResponseHandler: (0, import_provider_utils3.createJsonResponseHandler)(
         openAIChatResponseSchema
@@ -401,7 +529,7 @@ var OpenAIChatLanguageModel = class {
       abortSignal: options.abortSignal,
       fetch: this.config.fetch
     });
-    const { messages: rawPrompt, ...rawSettings } = args;
+    const { messages: rawPrompt, ...rawSettings } = body;
     const choice = response.choices[0];
     let providerMetadata;
     if (((_b = (_a = response.usage) == null ? void 0 : _a.completion_tokens_details) == null ? void 0 : _b.reasoning_tokens) != null || ((_d = (_c = response.usage) == null ? void 0 : _c.prompt_tokens_details) == null ? void 0 : _d.cached_tokens) != null) {
@@ -438,6 +566,7 @@ var OpenAIChatLanguageModel = class {
       },
       rawCall: { rawPrompt, rawSettings },
       rawResponse: { headers: responseHeaders },
+      request: { body: JSON.stringify(body) },
       response: getResponseMetadata(response),
       warnings,
       logprobs: mapOpenAIChatLogProbsOutput(choice.logprobs),
@@ -482,18 +611,19 @@ var OpenAIChatLanguageModel = class {
       };
     }
     const { args, warnings } = this.getArgs(options);
+    const body = {
+      ...args,
+      stream: true,
+      // only include stream_options when in strict compatibility mode:
+      stream_options: this.config.compatibility === "strict" ? { include_usage: true } : void 0
+    };
     const { responseHeaders, value: response } = await (0, import_provider_utils3.postJsonToApi)({
       url: this.config.url({
         path: "/chat/completions",
         modelId: this.modelId
       }),
       headers: (0, import_provider_utils3.combineHeaders)(this.config.headers(), options.headers),
-      body: {
-        ...args,
-        stream: true,
-        // only include stream_options when in strict compatibility mode:
-        stream_options: this.config.compatibility === "strict" ? { include_usage: true } : void 0
-      },
+      body,
       failedResponseHandler: openaiFailedResponseHandler,
       successfulResponseHandler: (0, import_provider_utils3.createEventSourceResponseHandler)(
         openaiChatChunkSchema
@@ -582,19 +712,19 @@ var OpenAIChatLanguageModel = class {
                 const index = toolCallDelta.index;
                 if (toolCalls[index] == null) {
                   if (toolCallDelta.type !== "function") {
-                    throw new import_provider2.InvalidResponseDataError({
+                    throw new import_provider3.InvalidResponseDataError({
                       data: toolCallDelta,
                       message: `Expected 'function' type.`
                     });
                   }
                   if (toolCallDelta.id == null) {
-                    throw new import_provider2.InvalidResponseDataError({
+                    throw new import_provider3.InvalidResponseDataError({
                       data: toolCallDelta,
                       message: `Expected 'id' to be a string.`
                     });
                   }
                   if (((_e = toolCallDelta.function) == null ? void 0 : _e.name) == null) {
-                    throw new import_provider2.InvalidResponseDataError({
+                    throw new import_provider3.InvalidResponseDataError({
                       data: toolCallDelta,
                       message: `Expected 'function.name' to be a string.`
                     });
@@ -670,6 +800,7 @@ var OpenAIChatLanguageModel = class {
       ),
       rawCall: { rawPrompt, rawSettings },
       rawResponse: { headers: responseHeaders },
+      request: { body: JSON.stringify(body) },
       warnings
     };
   }
@@ -776,91 +907,20 @@ var openaiChatChunkSchema = import_zod2.z.union([
   }),
   openAIErrorDataSchema
 ]);
-function prepareToolsAndToolChoice({
-  mode,
-  useLegacyFunctionCalling = false,
-  structuredOutputs = false
-}) {
-  var _a;
-  const tools = ((_a = mode.tools) == null ? void 0 : _a.length) ? mode.tools : void 0;
-  if (tools == null) {
-    return { tools: void 0, tool_choice: void 0 };
-  }
-  const toolChoice = mode.toolChoice;
-  if (useLegacyFunctionCalling) {
-    const mappedFunctions = tools.map((tool) => ({
-      name: tool.name,
-      description: tool.description,
-      parameters: tool.parameters
-    }));
-    if (toolChoice == null) {
-      return { functions: mappedFunctions, function_call: void 0 };
-    }
-    const type2 = toolChoice.type;
-    switch (type2) {
-      case "auto":
-      case "none":
-      case void 0:
-        return {
-          functions: mappedFunctions,
-          function_call: void 0
-        };
-      case "required":
-        throw new import_provider2.UnsupportedFunctionalityError({
-          functionality: "useLegacyFunctionCalling and toolChoice: required"
-        });
-      default:
-        return {
-          functions: mappedFunctions,
-          function_call: { name: toolChoice.toolName }
-        };
-    }
-  }
-  const mappedTools = tools.map((tool) => ({
-    type: "function",
-    function: {
-      name: tool.name,
-      description: tool.description,
-      parameters: tool.parameters,
-      strict: structuredOutputs === true ? true : void 0
-    }
-  }));
-  if (toolChoice == null) {
-    return { tools: mappedTools, tool_choice: void 0 };
-  }
-  const type = toolChoice.type;
-  switch (type) {
-    case "auto":
-    case "none":
-    case "required":
-      return { tools: mappedTools, tool_choice: type };
-    case "tool":
-      return {
-        tools: mappedTools,
-        tool_choice: {
-          type: "function",
-          function: {
-            name: toolChoice.toolName
-          }
-        }
-      };
-    default: {
-      const _exhaustiveCheck = type;
-      throw new Error(`Unsupported tool choice type: ${_exhaustiveCheck}`);
-    }
-  }
-}
 function isReasoningModel(modelId) {
   return modelId.startsWith("o1-");
 }
+function isAudioModel(modelId) {
+  return modelId.startsWith("gpt-4o-audio-preview");
+}
 
 // src/openai-completion-language-model.ts
-var import_provider4 = require("@ai-sdk/provider");
+var import_provider5 = require("@ai-sdk/provider");
 var import_provider_utils4 = require("@ai-sdk/provider-utils");
 var import_zod3 = require("zod");
 
 // src/convert-to-openai-completion-prompt.ts
-var import_provider3 = require("@ai-sdk/provider");
+var import_provider4 = require("@ai-sdk/provider");
 function convertToOpenAICompletionPrompt({
   prompt,
   inputFormat,
@@ -880,7 +940,7 @@ function convertToOpenAICompletionPrompt({
   for (const { role, content } of prompt) {
     switch (role) {
       case "system": {
-        throw new import_provider3.InvalidPromptError({
+        throw new import_provider4.InvalidPromptError({
           message: "Unexpected system message in prompt: ${content}",
           prompt
         });
@@ -892,7 +952,7 @@ function convertToOpenAICompletionPrompt({
               return part.text;
             }
             case "image": {
-              throw new import_provider3.UnsupportedFunctionalityError({
+              throw new import_provider4.UnsupportedFunctionalityError({
                 functionality: "images"
               });
             }
@@ -911,7 +971,7 @@ ${userMessage}
               return part.text;
             }
             case "tool-call": {
-              throw new import_provider3.UnsupportedFunctionalityError({
+              throw new import_provider4.UnsupportedFunctionalityError({
                 functionality: "tool-call messages"
               });
             }
@@ -924,7 +984,7 @@ ${assistantMessage}
         break;
       }
       case "tool": {
-        throw new import_provider3.UnsupportedFunctionalityError({
+        throw new import_provider4.UnsupportedFunctionalityError({
           functionality: "tool messages"
         });
       }
@@ -1025,24 +1085,24 @@ var OpenAICompletionLanguageModel = class {
     switch (type) {
       case "regular": {
         if ((_a = mode.tools) == null ? void 0 : _a.length) {
-          throw new import_provider4.UnsupportedFunctionalityError({
+          throw new import_provider5.UnsupportedFunctionalityError({
             functionality: "tools"
           });
         }
         if (mode.toolChoice) {
-          throw new import_provider4.UnsupportedFunctionalityError({
+          throw new import_provider5.UnsupportedFunctionalityError({
             functionality: "toolChoice"
           });
         }
         return { args: baseArgs, warnings };
       }
       case "object-json": {
-        throw new import_provider4.UnsupportedFunctionalityError({
+        throw new import_provider5.UnsupportedFunctionalityError({
           functionality: "object-json mode"
         });
       }
       case "object-tool": {
-        throw new import_provider4.UnsupportedFunctionalityError({
+        throw new import_provider5.UnsupportedFunctionalityError({
           functionality: "object-tool mode"
         });
       }
@@ -1081,23 +1141,25 @@ var OpenAICompletionLanguageModel = class {
       rawCall: { rawPrompt, rawSettings },
       rawResponse: { headers: responseHeaders },
       response: getResponseMetadata(response),
-      warnings
+      warnings,
+      request: { body: JSON.stringify(args) }
     };
   }
   async doStream(options) {
     const { args, warnings } = this.getArgs(options);
+    const body = {
+      ...args,
+      stream: true,
+      // only include stream_options when in strict compatibility mode:
+      stream_options: this.config.compatibility === "strict" ? { include_usage: true } : void 0
+    };
     const { responseHeaders, value: response } = await (0, import_provider_utils4.postJsonToApi)({
       url: this.config.url({
         path: "/completions",
         modelId: this.modelId
       }),
       headers: (0, import_provider_utils4.combineHeaders)(this.config.headers(), options.headers),
-      body: {
-        ...args,
-        stream: true,
-        // only include stream_options when in strict compatibility mode:
-        stream_options: this.config.compatibility === "strict" ? { include_usage: true } : void 0
-      },
+      body,
       failedResponseHandler: openaiFailedResponseHandler,
       successfulResponseHandler: (0, import_provider_utils4.createEventSourceResponseHandler)(
         openaiCompletionChunkSchema
@@ -1171,7 +1233,8 @@ var OpenAICompletionLanguageModel = class {
       ),
       rawCall: { rawPrompt, rawSettings },
       rawResponse: { headers: responseHeaders },
-      warnings
+      warnings,
+      request: { body: JSON.stringify(body) }
     };
   }
 };
@@ -1271,7 +1334,7 @@ var OpenAI = class {
 var import_provider_utils7 = require("@ai-sdk/provider-utils");
 
 // src/openai-embedding-model.ts
-var import_provider5 = require("@ai-sdk/provider");
+var import_provider6 = require("@ai-sdk/provider");
 var import_provider_utils6 = require("@ai-sdk/provider-utils");
 var import_zod4 = require("zod");
 var OpenAIEmbeddingModel = class {
@@ -1298,7 +1361,7 @@ var OpenAIEmbeddingModel = class {
     abortSignal
   }) {
     if (values.length > this.maxEmbeddingsPerCall) {
-      throw new import_provider5.TooManyEmbeddingValuesForCallError({
+      throw new import_provider6.TooManyEmbeddingValuesForCallError({
         provider: this.provider,
         modelId: this.modelId,
         maxEmbeddingsPerCall: this.maxEmbeddingsPerCall,