@llmgateway/ai-sdk-provider 3.7.0 → 3.8.1

package/dist/internal/index.d.mts

@@ -1,6 +1,5 @@
 import { LanguageModelV3, LanguageModelV3CallOptions, ImageModelV3, ImageModelV3CallOptions, SharedV3Warning, LanguageModelV3GenerateResult, LanguageModelV3StreamResult } from '@ai-sdk/provider';
 export { LanguageModelV3, LanguageModelV3Prompt } from '@ai-sdk/provider';
-import { models, Provider } from '@llmgateway/models';
 export { Model, ModelDefinition, Provider, ProviderModelMapping, StabilityLevel, models } from '@llmgateway/models';
 
 type LLMGatewayProviderOptions = {
@@ -154,6 +153,70 @@ declare class LLMGatewayImageModel implements ImageModelV3 {
     }>;
 }
 
+type LLMGatewayChatModelId = string;
+type LLMGatewayChatSettings = {
+    /**
+  Modify the likelihood of specified tokens appearing in the completion.
+  
+  Accepts a JSON object that maps tokens (specified by their token ID in
+  the GPT tokenizer) to an associated bias value from -100 to 100. You
+  can use this tokenizer tool to convert text to token IDs. Mathematically,
+  the bias is added to the logits generated by the model prior to sampling.
+  The exact effect will vary per model, but values between -1 and 1 should
+  decrease or increase likelihood of selection; values like -100 or 100
+  should result in a ban or exclusive selection of the relevant token.
+  
+  As an example, you can pass {"50256": -100} to prevent the <|endoftext|>
+  token from being generated.
+  */
+    logitBias?: Record<number, number>;
+    /**
+  Return the log probabilities of the tokens. Including logprobs will increase
+  the response size and can slow down response times. However, it can
+  be useful to understand better how the model is behaving.
+  
+  Setting to true will return the log probabilities of the tokens that
+  were generated.
+  
+  Setting to a number will return the log probabilities of the top n
+  tokens that were generated.
+  */
+    logprobs?: boolean | number;
+    /**
+  Whether to enable parallel function calling during tool use. Default to true.
+     */
+    parallelToolCalls?: boolean;
+    /**
+  A unique identifier representing your end-user, which can help LLMGateway to
+  monitor and detect abuse. Learn more.
+  */
+    user?: string;
+} & LLMGatewaySharedSettings;
+
+type LLMGatewayChatConfig = {
+    provider: string;
+    compatibility: 'strict' | 'compatible';
+    headers: () => Record<string, string | undefined>;
+    url: (options: {
+        modelId: string;
+        path: string;
+    }) => string;
+    fetch?: typeof fetch;
+    extraBody?: Record<string, unknown>;
+};
+declare class LLMGatewayChatLanguageModel implements LanguageModelV3 {
+    readonly specificationVersion: "v3";
+    readonly provider = "llmgateway";
+    readonly modelId: LLMGatewayChatModelId;
+    readonly supportedUrls: Record<string, RegExp[]>;
+    readonly settings: LLMGatewayChatSettings;
+    private readonly config;
+    constructor(modelId: LLMGatewayChatModelId, settings: LLMGatewayChatSettings, config: LLMGatewayChatConfig);
+    private getArgs;
+    doGenerate(options: LanguageModelV3CallOptions): Promise<LanguageModelV3GenerateResult>;
+    doStream(options: LanguageModelV3CallOptions): Promise<LanguageModelV3StreamResult>;
+}
+
 declare const providers: [{
     readonly id: "llmgateway";
     readonly name: "LLM Gateway";
@@ -343,68 +406,4 @@ declare const providers: [{
 }];
 type ProviderId = (typeof providers)[number]['id'];
 
-type LLMGatewayChatModelId = (typeof models)[number]['id'] | `${Provider}/${(typeof models)[number]['id']}` | 'test-model';
-type LLMGatewayChatSettings = {
-    /**
-  Modify the likelihood of specified tokens appearing in the completion.
-  
-  Accepts a JSON object that maps tokens (specified by their token ID in
-  the GPT tokenizer) to an associated bias value from -100 to 100. You
-  can use this tokenizer tool to convert text to token IDs. Mathematically,
-  the bias is added to the logits generated by the model prior to sampling.
-  The exact effect will vary per model, but values between -1 and 1 should
-  decrease or increase likelihood of selection; values like -100 or 100
-  should result in a ban or exclusive selection of the relevant token.
-  
-  As an example, you can pass {"50256": -100} to prevent the <|endoftext|>
-  token from being generated.
-  */
-    logitBias?: Record<number, number>;
-    /**
-  Return the log probabilities of the tokens. Including logprobs will increase
-  the response size and can slow down response times. However, it can
-  be useful to understand better how the model is behaving.
-  
-  Setting to true will return the log probabilities of the tokens that
-  were generated.
-  
-  Setting to a number will return the log probabilities of the top n
-  tokens that were generated.
-  */
-    logprobs?: boolean | number;
-    /**
-  Whether to enable parallel function calling during tool use. Default to true.
-     */
-    parallelToolCalls?: boolean;
-    /**
-  A unique identifier representing your end-user, which can help LLMGateway to
-  monitor and detect abuse. Learn more.
-  */
-    user?: string;
-} & LLMGatewaySharedSettings;
-
-type LLMGatewayChatConfig = {
-    provider: string;
-    compatibility: 'strict' | 'compatible';
-    headers: () => Record<string, string | undefined>;
-    url: (options: {
-        modelId: string;
-        path: string;
-    }) => string;
-    fetch?: typeof fetch;
-    extraBody?: Record<string, unknown>;
-};
-declare class LLMGatewayChatLanguageModel implements LanguageModelV3 {
-    readonly specificationVersion: "v3";
-    readonly provider = "llmgateway";
-    readonly modelId: LLMGatewayChatModelId;
-    readonly supportedUrls: Record<string, RegExp[]>;
-    readonly settings: LLMGatewayChatSettings;
-    private readonly config;
-    constructor(modelId: LLMGatewayChatModelId, settings: LLMGatewayChatSettings, config: LLMGatewayChatConfig);
-    private getArgs;
-    doGenerate(options: LanguageModelV3CallOptions): Promise<LanguageModelV3GenerateResult>;
-    doStream(options: LanguageModelV3CallOptions): Promise<LanguageModelV3StreamResult>;
-}
-
 export { LLMGatewayChatLanguageModel, type LLMGatewayChatModelId, type LLMGatewayChatSettings, LLMGatewayCompletionLanguageModel, type LLMGatewayCompletionSettings, LLMGatewayImageModel, type LLMGatewayImageModelId, type LLMGatewayImageSettings, type LLMGatewayProviderOptions, type LLMGatewaySharedSettings, type LLMGatewayUsageAccounting, type ProviderId, providers };

package/dist/internal/index.d.ts

@@ -1,6 +1,5 @@
 import { LanguageModelV3, LanguageModelV3CallOptions, ImageModelV3, ImageModelV3CallOptions, SharedV3Warning, LanguageModelV3GenerateResult, LanguageModelV3StreamResult } from '@ai-sdk/provider';
 export { LanguageModelV3, LanguageModelV3Prompt } from '@ai-sdk/provider';
-import { models, Provider } from '@llmgateway/models';
 export { Model, ModelDefinition, Provider, ProviderModelMapping, StabilityLevel, models } from '@llmgateway/models';
 
 type LLMGatewayProviderOptions = {
@@ -154,6 +153,70 @@ declare class LLMGatewayImageModel implements ImageModelV3 {
     }>;
 }
 
+type LLMGatewayChatModelId = string;
+type LLMGatewayChatSettings = {
+    /**
+  Modify the likelihood of specified tokens appearing in the completion.
+  
+  Accepts a JSON object that maps tokens (specified by their token ID in
+  the GPT tokenizer) to an associated bias value from -100 to 100. You
+  can use this tokenizer tool to convert text to token IDs. Mathematically,
+  the bias is added to the logits generated by the model prior to sampling.
+  The exact effect will vary per model, but values between -1 and 1 should
+  decrease or increase likelihood of selection; values like -100 or 100
+  should result in a ban or exclusive selection of the relevant token.
+  
+  As an example, you can pass {"50256": -100} to prevent the <|endoftext|>
+  token from being generated.
+  */
+    logitBias?: Record<number, number>;
+    /**
+  Return the log probabilities of the tokens. Including logprobs will increase
+  the response size and can slow down response times. However, it can
+  be useful to understand better how the model is behaving.
+  
+  Setting to true will return the log probabilities of the tokens that
+  were generated.
+  
+  Setting to a number will return the log probabilities of the top n
+  tokens that were generated.
+  */
+    logprobs?: boolean | number;
+    /**
+  Whether to enable parallel function calling during tool use. Default to true.
+     */
+    parallelToolCalls?: boolean;
+    /**
+  A unique identifier representing your end-user, which can help LLMGateway to
+  monitor and detect abuse. Learn more.
+  */
+    user?: string;
+} & LLMGatewaySharedSettings;
+
+type LLMGatewayChatConfig = {
+    provider: string;
+    compatibility: 'strict' | 'compatible';
+    headers: () => Record<string, string | undefined>;
+    url: (options: {
+        modelId: string;
+        path: string;
+    }) => string;
+    fetch?: typeof fetch;
+    extraBody?: Record<string, unknown>;
+};
+declare class LLMGatewayChatLanguageModel implements LanguageModelV3 {
+    readonly specificationVersion: "v3";
+    readonly provider = "llmgateway";
+    readonly modelId: LLMGatewayChatModelId;
+    readonly supportedUrls: Record<string, RegExp[]>;
+    readonly settings: LLMGatewayChatSettings;
+    private readonly config;
+    constructor(modelId: LLMGatewayChatModelId, settings: LLMGatewayChatSettings, config: LLMGatewayChatConfig);
+    private getArgs;
+    doGenerate(options: LanguageModelV3CallOptions): Promise<LanguageModelV3GenerateResult>;
+    doStream(options: LanguageModelV3CallOptions): Promise<LanguageModelV3StreamResult>;
+}
+
 declare const providers: [{
     readonly id: "llmgateway";
     readonly name: "LLM Gateway";
@@ -343,68 +406,4 @@ declare const providers: [{
 }];
 type ProviderId = (typeof providers)[number]['id'];
 
-type LLMGatewayChatModelId = (typeof models)[number]['id'] | `${Provider}/${(typeof models)[number]['id']}` | 'test-model';
-type LLMGatewayChatSettings = {
-    /**
-  Modify the likelihood of specified tokens appearing in the completion.
-  
-  Accepts a JSON object that maps tokens (specified by their token ID in
-  the GPT tokenizer) to an associated bias value from -100 to 100. You
-  can use this tokenizer tool to convert text to token IDs. Mathematically,
-  the bias is added to the logits generated by the model prior to sampling.
-  The exact effect will vary per model, but values between -1 and 1 should
-  decrease or increase likelihood of selection; values like -100 or 100
-  should result in a ban or exclusive selection of the relevant token.
-  
-  As an example, you can pass {"50256": -100} to prevent the <|endoftext|>
-  token from being generated.
-  */
-    logitBias?: Record<number, number>;
-    /**
-  Return the log probabilities of the tokens. Including logprobs will increase
-  the response size and can slow down response times. However, it can
-  be useful to understand better how the model is behaving.
-  
-  Setting to true will return the log probabilities of the tokens that
-  were generated.
-  
-  Setting to a number will return the log probabilities of the top n
-  tokens that were generated.
-  */
-    logprobs?: boolean | number;
-    /**
-  Whether to enable parallel function calling during tool use. Default to true.
-     */
-    parallelToolCalls?: boolean;
-    /**
-  A unique identifier representing your end-user, which can help LLMGateway to
-  monitor and detect abuse. Learn more.
-  */
-    user?: string;
-} & LLMGatewaySharedSettings;
-
-type LLMGatewayChatConfig = {
-    provider: string;
-    compatibility: 'strict' | 'compatible';
-    headers: () => Record<string, string | undefined>;
-    url: (options: {
-        modelId: string;
-        path: string;
-    }) => string;
-    fetch?: typeof fetch;
-    extraBody?: Record<string, unknown>;
-};
-declare class LLMGatewayChatLanguageModel implements LanguageModelV3 {
-    readonly specificationVersion: "v3";
-    readonly provider = "llmgateway";
-    readonly modelId: LLMGatewayChatModelId;
-    readonly supportedUrls: Record<string, RegExp[]>;
-    readonly settings: LLMGatewayChatSettings;
-    private readonly config;
-    constructor(modelId: LLMGatewayChatModelId, settings: LLMGatewayChatSettings, config: LLMGatewayChatConfig);
-    private getArgs;
-    doGenerate(options: LanguageModelV3CallOptions): Promise<LanguageModelV3GenerateResult>;
-    doStream(options: LanguageModelV3CallOptions): Promise<LanguageModelV3StreamResult>;
-}
-
 export { LLMGatewayChatLanguageModel, type LLMGatewayChatModelId, type LLMGatewayChatSettings, LLMGatewayCompletionLanguageModel, type LLMGatewayCompletionSettings, LLMGatewayImageModel, type LLMGatewayImageModelId, type LLMGatewayImageSettings, type LLMGatewayProviderOptions, type LLMGatewaySharedSettings, type LLMGatewayUsageAccounting, type ProviderId, providers };

package/dist/internal/index.js

@@ -2275,6 +2275,29 @@ function getBase64FromDataUrl(dataUrl) {
 }
 
 // src/chat/convert-to-llmgateway-chat-messages.ts
+var AUDIO_FORMAT_BY_MIME = {
+  "audio/wav": "wav",
+  "audio/wave": "wav",
+  "audio/x-wav": "wav",
+  "audio/mpeg": "mp3",
+  "audio/mp3": "mp3",
+  "audio/mp4": "mp4",
+  "audio/m4a": "m4a",
+  "audio/x-m4a": "m4a",
+  "audio/aac": "aac",
+  "audio/ogg": "ogg",
+  "audio/flac": "flac",
+  "audio/aiff": "aiff",
+  "audio/x-aiff": "aiff",
+  "audio/webm": "webm"
+};
+function audioFormatFromMime(mime) {
+  var _a16;
+  if (!mime) return void 0;
+  const lower = (_a16 = mime.toLowerCase().split(";")[0]) == null ? void 0 : _a16.trim();
+  if (!lower) return void 0;
+  return AUDIO_FORMAT_BY_MIME[lower];
+}
 function getCacheControl(providerOptions) {
   var _a16, _b16, _c;
   const anthropic = providerOptions == null ? void 0 : providerOptions.anthropic;
@@ -2313,7 +2336,7 @@ function convertToLLMGatewayChatMessages(prompt) {
         const messageCacheControl = getCacheControl(providerOptions);
         const contentParts = content.map(
           (part) => {
-            var _a17, _b17, _c2, _d, _e, _f;
+            var _a17, _b17, _c2, _d, _e, _f, _g;
             const cacheControl = (_a17 = getCacheControl(part.providerOptions)) != null ? _a17 : messageCacheControl;
             switch (part.type) {
               case "text":
@@ -2338,8 +2361,27 @@ function convertToLLMGatewayChatMessages(prompt) {
                     cache_control: cacheControl
                   };
                 }
+                if ((_c2 = part.mediaType) == null ? void 0 : _c2.startsWith("audio/")) {
+                  const format = audioFormatFromMime(part.mediaType);
+                  if (format) {
+                    const fileUrl = getFileUrl({
+                      part,
+                      defaultMediaType: part.mediaType
+                    });
+                    if (fileUrl.startsWith("data:")) {
+                      return {
+                        type: "input_audio",
+                        input_audio: {
+                          data: getBase64FromDataUrl(fileUrl),
+                          format
+                        },
+                        cache_control: cacheControl
+                      };
+                    }
+                  }
+                }
                 const fileName = String(
-                  (_f = (_e = (_d = (_c2 = part.providerOptions) == null ? void 0 : _c2.llmgateway) == null ? void 0 : _d.filename) != null ? _e : part.filename) != null ? _f : ""
+                  (_g = (_f = (_e = (_d = part.providerOptions) == null ? void 0 : _d.llmgateway) == null ? void 0 : _e.filename) != null ? _f : part.filename) != null ? _g : ""
                 );
                 const fileData = getFileUrl({
                   part,