@aihubmix/ai-sdk-provider 2.0.4 → 2.0.6

package/CHANGELOG.md

@@ -1,34 +1,10 @@
 # @aihubmix/ai-sdk-provider
 
-## 2.0.2
+## 2.0.6
 
 ### Patch Changes
 
-- b6c5f35: - 将 Zod 降级到 ^3.25.76 以保证与 AI SDK 生态系统的稳定性。
-  - 移除调试日志。
-- 7a9e37e: - 添加 `aihubmixChatProviderOptionsSchema` 和 `aihubmixResponsesProviderOptionsSchema` 以支持所有 Vercel AI SDK OpenAI 提供商选项。
-  - 从主入口点导出选项类型和模式。
-  - 改进了 `gpt-5-pro` 和 `gpt-5-codex` 的模型配置。
-  - 通过更好的文件命名逻辑增强了 `AihubmixTranscriptionModel`。
-
-## 2.0.0
-
-### Major Changes
-
-- 2dc97a3: 升级到 AI SDK v6
-
-  - 升级所有 AI SDK 依赖到 v6 兼容版本
-  - Provider 接口从 V2 升级到 V3 (LanguageModelV3, ProviderV3 等)
-  - 工具工厂函数 API 更新 (createProviderToolFactory)
-  - 添加 specificationVersion 和 embeddingModel 属性
-  - 版本号升级到 1.0.0
-
-### Patch Changes
-
-- f3b2c86: - 添加 `aihubmixChatProviderOptionsSchema` 和 `aihubmixResponsesProviderOptionsSchema` 以支持所有 Vercel AI SDK OpenAI 提供商选项。
-  - 从主入口点导出选项类型和模式。
-  - 改进了 `gpt-5-pro` 和 `gpt-5-codex` 的模型配置。
-  - 通过更好的文件命名逻辑增强了 `AihubmixTranscriptionModel`。
+- 修复 Anthropic / Google 模型透传 `Authorization` 头的问题:这两个 SDK 自带 `x-api-key` / `x-goog-api-key`,不再额外发送 `Authorization`,避免上游对冗余鉴权头敏感。
 
 ## 1.0.1
 

package/README.ja.md

@@ -14,7 +14,17 @@ app-codeが内蔵されており、この方法でモデルをリクエストす
 [AI SDK](https://ai-sdk.dev/docs)用の **[Aihubmix provider](https://sdk.vercel.ai/providers/community-providers/aihubmix)**
 一つのゲートウェイ、無限のモデル；ワンストップリクエスト：OpenAI、Claude、Gemini、DeepSeek、Qwen、そして500以上のAIモデル。
 
-> **📦 バージョン 1.0.1** - AI SDK v6 対応
+## AI SDK v6 用セットアップ
+
+```bash
+npm i @aihubmix/ai-sdk-provider
+```
+
+### （レガシー）AI SDK v5 用セットアップ
+
+```bash
+npm i @aihubmix/ai-sdk-provider@0.0.6
+```
 
 ## サポートされている機能
 
@@ -30,14 +40,6 @@ Aihubmix providerは以下のAI機能をサポートしています：
 - **転写**：音声からテキストへの変換
 - **ツール**：ウェブ検索およびその他のツール
 
-## セットアップ
-
-Aihubmix providerは`@aihubmix/ai-sdk-provider`モジュールで利用可能です。[@aihubmix/ai-sdk-provider](https://www.npmjs.com/package/@aihubmix/ai-sdk-provider)でインストールできます
-
-```bash
-npm i @aihubmix/ai-sdk-provider
-```
-
 ## Provider インスタンス
 
 `@aihubmix/ai-sdk-provider`からデフォルトのproviderインスタンス`aihubmix`をインポートできます：
@@ -299,7 +301,7 @@ const { text } = await generateText({
 
 ## 追加リソース
 
-- [Aihubmix Provider リポジトリ](https://github.com/inferera/aihubmix)
+- [Aihubmix Provider リポジトリ](https://github.com/AIhubmix/ai-sdk-provider)
 - [Aihubmix ドキュメント](https://docs.aihubmix.com/en)
 - [Aihubmix ダッシュボード](https://aihubmix.com)
 - [Aihubmix ビジネス協力](mailto:business@aihubmix.com)

package/README.md

@@ -14,8 +14,17 @@ Built-in app-code; using this method to request all models offers a 10% discount
 The **[Aihubmix provider](https://sdk.vercel.ai/providers/community-providers/aihubmix)** for the [AI SDK](https://ai-sdk.dev/docs)
 One Gateway, Infinite Models；one-stop request: OpenAI, Claude, Gemini, DeepSeek, Qwen, and over 500 AI models.
 
-> **📦 Version 1.0.1** - Compatible with AI SDK v6
+## Setup for AI SDK v6
 
+```bash
+npm i @aihubmix/ai-sdk-provider
+```
+
+### (LEGACY) Setup for AI SDK v5
+
+```bash
+npm i @aihubmix/ai-sdk-provider@0.0.6
+```
 
 ## Supported Features
 
@@ -32,14 +41,6 @@ The Aihubmix provider supports the following AI features:
 - **Tools**: Web search and other tools
 
 
-## Setup
-
-The Aihubmix provider is available in the `@aihubmix/ai-sdk-provider` module. You can install it with [@aihubmix/ai-sdk-provider](https://www.npmjs.com/package/@aihubmix/ai-sdk-provider)
-
-```bash
-npm i @aihubmix/ai-sdk-provider
-```
-
 ## Provider Instance
 
 You can import the default provider instance `aihubmix` from `@aihubmix/ai-sdk-provider`:
@@ -302,7 +303,7 @@ const { text } = await generateText({
 
 ## Additional Resources
 
-- [Aihubmix Provider Repository](https://github.com/inferera/aihubmix)
+- [Aihubmix Provider Repository](https://github.com/AIhubmix/ai-sdk-provider)
 - [Aihubmix Documentation](https://docs.aihubmix.com/en)
 - [Aihubmix Dashboard](https://aihubmix.com)
 - [Aihubmix Cooperation](mailto:business@aihubmix.com)

package/README.zh.md

@@ -14,7 +14,17 @@
 **[Aihubmix provider](https://sdk.vercel.ai/providers/community-providers/aihubmix)** 适用于 [AI SDK](https://ai-sdk.dev/docs)
 一个网关，无限模型；一站式请求：OpenAI、Claude、Gemini、DeepSeek、Qwen 以及超过 500 个 AI 模型。
 
-> **📦 版本 1.0.1** - 兼容 AI SDK v6
+## 安装 AI SDK v6 版本
+
+```bash
+npm i @aihubmix/ai-sdk-provider
+```
+
+### （旧版）安装 AI SDK v5 版本
+
+```bash
+npm i @aihubmix/ai-sdk-provider@0.0.6
+```
 
 ## 支持的功能
 
@@ -30,14 +40,6 @@ Aihubmix provider 支持以下 AI 功能：
 - **转录**：语音转文本转换
 - **工具**：网络搜索和其他工具
 
-## 安装
-
-Aihubmix 在 `@aihubmix/ai-sdk-provider` 模块中可用。您可以通过 [@aihubmix/ai-sdk-provider](https://www.npmjs.com/package/@aihubmix/ai-sdk-provider) 安装它
-
-```bash
-npm i @aihubmix/ai-sdk-provider
-```
-
 ## Provider 实例
 
 您可以从 `@aihubmix/ai-sdk-provider` 导入默认的 provider 实例 `aihubmix`：
@@ -299,7 +301,7 @@ const { text } = await generateText({
 
 ## 附加资源
 
-- [Aihubmix Provider 仓库](https://github.com/inferera/aihubmix)
+- [Aihubmix Provider 仓库](https://github.com/AIhubmix/ai-sdk-provider)
 - [Aihubmix 文档](https://docs.aihubmix.com/en)
 - [Aihubmix 控制台](https://aihubmix.com)
 - [Aihubmix 商务合作](mailto:business@aihubmix.com) 

package/dist/index.d.mts

@@ -1,6 +1,6 @@
-import * as _ai_sdk_provider_utils from '@ai-sdk/provider-utils';
-import { InferSchema, FetchFunction } from '@ai-sdk/provider-utils';
 import { ProviderV3, LanguageModelV3, EmbeddingModelV3, ImageModelV3, TranscriptionModelV3, SpeechModelV3 } from '@ai-sdk/provider';
+import * as _ai_sdk_provider_utils from '@ai-sdk/provider-utils';
+import { FetchFunction } from '@ai-sdk/provider-utils';
 
 declare const webSearchToolFactory: _ai_sdk_provider_utils.ProviderToolFactory<{}, {
     /**
@@ -198,58 +198,9 @@ declare const aihubmixTools: {
     webSearch: (args?: Parameters<typeof webSearchToolFactory>[0]) => _ai_sdk_provider_utils.Tool<{}, unknown>;
 };
 
-/**
- * OpenAI Chat 语言模型的 Provider Options Schema
- * 与 Vercel AI SDK 的 openaiChatLanguageModelOptions 保持一致
- */
-declare const aihubmixChatProviderOptionsSchema: _ai_sdk_provider_utils.LazySchema<{
-    logitBias?: Record<string, number> | undefined;
-    logprobs?: number | boolean | undefined;
-    parallelToolCalls?: boolean | undefined;
-    user?: string | undefined;
-    reasoningEffort?: "low" | "high" | "medium" | "none" | "minimal" | "xhigh" | undefined;
-    maxCompletionTokens?: number | undefined;
-    store?: boolean | undefined;
-    metadata?: Record<string, string> | undefined;
-    prediction?: Record<string, any> | undefined;
-    serviceTier?: "auto" | "flex" | "priority" | "default" | undefined;
-    strictJsonSchema?: boolean | undefined;
-    textVerbosity?: "low" | "high" | "medium" | undefined;
-    promptCacheKey?: string | undefined;
-    promptCacheRetention?: "in_memory" | "24h" | undefined;
-    safetyIdentifier?: string | undefined;
-    systemMessageMode?: "system" | "developer" | "remove" | undefined;
-    forceReasoning?: boolean | undefined;
-}>;
-/**
- * OpenAI Responses API 的 Provider Options Schema
- */
-declare const aihubmixResponsesProviderOptionsSchema: _ai_sdk_provider_utils.LazySchema<{
-    logprobs?: number | boolean | undefined;
-    parallelToolCalls?: boolean | null | undefined;
-    user?: string | null | undefined;
-    reasoningEffort?: string | null | undefined;
-    store?: boolean | null | undefined;
-    metadata?: any;
-    serviceTier?: "auto" | "flex" | "priority" | "default" | null | undefined;
-    strictJsonSchema?: boolean | null | undefined;
-    textVerbosity?: "low" | "high" | "medium" | null | undefined;
-    promptCacheKey?: string | null | undefined;
-    promptCacheRetention?: "in_memory" | "24h" | null | undefined;
-    safetyIdentifier?: string | null | undefined;
-    systemMessageMode?: "system" | "developer" | "remove" | undefined;
-    forceReasoning?: boolean | undefined;
-    conversation?: string | null | undefined;
-    include?: ("reasoning.encrypted_content" | "file_search_call.results" | "message.output_text.logprobs")[] | null | undefined;
-    instructions?: string | null | undefined;
-    maxToolCalls?: number | null | undefined;
-    previousResponseId?: string | null | undefined;
-    reasoningSummary?: string | null | undefined;
-    truncation?: "auto" | "disabled" | null | undefined;
-}>;
-type AihubmixChatProviderOptions = InferSchema<typeof aihubmixChatProviderOptionsSchema>;
-type AihubmixResponsesProviderOptions = InferSchema<typeof aihubmixResponsesProviderOptionsSchema>;
-type OpenAIProviderSettings = AihubmixChatProviderOptions;
+interface OpenAIProviderSettings {
+    [key: string]: unknown;
+}
 interface AihubmixProvider extends ProviderV3 {
     (deploymentId: string, settings?: OpenAIProviderSettings): LanguageModelV3;
     readonly specificationVersion: 'v3';
@@ -276,4 +227,4 @@ interface AihubmixProviderSettings {
 declare function createAihubmix(options?: AihubmixProviderSettings): AihubmixProvider;
 declare const aihubmix: AihubmixProvider;
 
-export { type AihubmixChatProviderOptions, type AihubmixProvider, type AihubmixProviderSettings, type AihubmixResponsesProviderOptions, aihubmix, aihubmixChatProviderOptionsSchema, aihubmixResponsesProviderOptionsSchema, createAihubmix };
+export { type AihubmixProvider, type AihubmixProviderSettings, aihubmix, createAihubmix };

package/dist/index.d.ts

@@ -1,6 +1,6 @@
-import * as _ai_sdk_provider_utils from '@ai-sdk/provider-utils';
-import { InferSchema, FetchFunction } from '@ai-sdk/provider-utils';
 import { ProviderV3, LanguageModelV3, EmbeddingModelV3, ImageModelV3, TranscriptionModelV3, SpeechModelV3 } from '@ai-sdk/provider';
+import * as _ai_sdk_provider_utils from '@ai-sdk/provider-utils';
+import { FetchFunction } from '@ai-sdk/provider-utils';
 
 declare const webSearchToolFactory: _ai_sdk_provider_utils.ProviderToolFactory<{}, {
     /**
@@ -198,58 +198,9 @@ declare const aihubmixTools: {
     webSearch: (args?: Parameters<typeof webSearchToolFactory>[0]) => _ai_sdk_provider_utils.Tool<{}, unknown>;
 };
 
-/**
- * OpenAI Chat 语言模型的 Provider Options Schema
- * 与 Vercel AI SDK 的 openaiChatLanguageModelOptions 保持一致
- */
-declare const aihubmixChatProviderOptionsSchema: _ai_sdk_provider_utils.LazySchema<{
-    logitBias?: Record<string, number> | undefined;
-    logprobs?: number | boolean | undefined;
-    parallelToolCalls?: boolean | undefined;
-    user?: string | undefined;
-    reasoningEffort?: "low" | "high" | "medium" | "none" | "minimal" | "xhigh" | undefined;
-    maxCompletionTokens?: number | undefined;
-    store?: boolean | undefined;
-    metadata?: Record<string, string> | undefined;
-    prediction?: Record<string, any> | undefined;
-    serviceTier?: "auto" | "flex" | "priority" | "default" | undefined;
-    strictJsonSchema?: boolean | undefined;
-    textVerbosity?: "low" | "high" | "medium" | undefined;
-    promptCacheKey?: string | undefined;
-    promptCacheRetention?: "in_memory" | "24h" | undefined;
-    safetyIdentifier?: string | undefined;
-    systemMessageMode?: "system" | "developer" | "remove" | undefined;
-    forceReasoning?: boolean | undefined;
-}>;
-/**
- * OpenAI Responses API 的 Provider Options Schema
- */
-declare const aihubmixResponsesProviderOptionsSchema: _ai_sdk_provider_utils.LazySchema<{
-    logprobs?: number | boolean | undefined;
-    parallelToolCalls?: boolean | null | undefined;
-    user?: string | null | undefined;
-    reasoningEffort?: string | null | undefined;
-    store?: boolean | null | undefined;
-    metadata?: any;
-    serviceTier?: "auto" | "flex" | "priority" | "default" | null | undefined;
-    strictJsonSchema?: boolean | null | undefined;
-    textVerbosity?: "low" | "high" | "medium" | null | undefined;
-    promptCacheKey?: string | null | undefined;
-    promptCacheRetention?: "in_memory" | "24h" | null | undefined;
-    safetyIdentifier?: string | null | undefined;
-    systemMessageMode?: "system" | "developer" | "remove" | undefined;
-    forceReasoning?: boolean | undefined;
-    conversation?: string | null | undefined;
-    include?: ("reasoning.encrypted_content" | "file_search_call.results" | "message.output_text.logprobs")[] | null | undefined;
-    instructions?: string | null | undefined;
-    maxToolCalls?: number | null | undefined;
-    previousResponseId?: string | null | undefined;
-    reasoningSummary?: string | null | undefined;
-    truncation?: "auto" | "disabled" | null | undefined;
-}>;
-type AihubmixChatProviderOptions = InferSchema<typeof aihubmixChatProviderOptionsSchema>;
-type AihubmixResponsesProviderOptions = InferSchema<typeof aihubmixResponsesProviderOptionsSchema>;
-type OpenAIProviderSettings = AihubmixChatProviderOptions;
+interface OpenAIProviderSettings {
+    [key: string]: unknown;
+}
 interface AihubmixProvider extends ProviderV3 {
     (deploymentId: string, settings?: OpenAIProviderSettings): LanguageModelV3;
     readonly specificationVersion: 'v3';
@@ -276,4 +227,4 @@ interface AihubmixProviderSettings {
 declare function createAihubmix(options?: AihubmixProviderSettings): AihubmixProvider;
 declare const aihubmix: AihubmixProvider;
 
-export { type AihubmixChatProviderOptions, type AihubmixProvider, type AihubmixProviderSettings, type AihubmixResponsesProviderOptions, aihubmix, aihubmixChatProviderOptionsSchema, aihubmixResponsesProviderOptionsSchema, createAihubmix };
+export { type AihubmixProvider, type AihubmixProviderSettings, aihubmix, createAihubmix };

package/dist/index.js

@@ -21,18 +21,16 @@ var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: tru
 var index_exports = {};
 __export(index_exports, {
   aihubmix: () => aihubmix,
-  aihubmixChatProviderOptionsSchema: () => aihubmixChatProviderOptionsSchema,
-  aihubmixResponsesProviderOptionsSchema: () => aihubmixResponsesProviderOptionsSchema,
   createAihubmix: () => createAihubmix
 });
 module.exports = __toCommonJS(index_exports);
 
 // src/aihubmix-provider.ts
+var import_openai_compatible = require("@ai-sdk/openai-compatible");
 var import_internal = require("@ai-sdk/openai/internal");
 var import_internal2 = require("@ai-sdk/anthropic/internal");
 var import_internal3 = require("@ai-sdk/google/internal");
 var import_provider_utils6 = require("@ai-sdk/provider-utils");
-var import_zod6 = require("zod");
 
 // src/tool/code-interpreter.ts
 var import_provider_utils = require("@ai-sdk/provider-utils");
@@ -299,191 +297,6 @@ var aihubmixTools = {
 };
 
 // src/aihubmix-provider.ts
-var aihubmixChatProviderOptionsSchema = (0, import_provider_utils6.lazySchema)(
-  () => (0, import_provider_utils6.zodSchema)(
-    import_zod6.z.object({
-      /**
-       * 修改指定 token 出现在生成内容中的概率
-       * 接受一个 JSON 对象，将 token ID（字符串形式）映射到 -100 到 100 之间的偏差值
-       */
-      logitBias: import_zod6.z.record(import_zod6.z.string(), import_zod6.z.number()).optional(),
-      /**
-       * 返回 token 的对数概率
-       * 设置为 true 返回生成 token 的对数概率
-       * 设置为数字返回前 n 个 token 的对数概率
-       */
-      logprobs: import_zod6.z.union([import_zod6.z.boolean(), import_zod6.z.number()]).optional(),
-      /**
-       * 是否启用并行工具调用，默认为 true
-       */
-      parallelToolCalls: import_zod6.z.boolean().optional(),
-      /**
-       * 代表终端用户的唯一标识符，帮助 OpenAI 监控和检测滥用行为
-       */
-      user: import_zod6.z.string().optional(),
-      /**
-       * 推理模型的推理强度，默认为 `medium`
-       */
-      reasoningEffort: import_zod6.z.enum(["none", "minimal", "low", "medium", "high", "xhigh"]).optional(),
-      /**
-       * 生成的最大完成 token 数，适用于推理模型
-       */
-      maxCompletionTokens: import_zod6.z.number().optional(),
-      /**
-       * 是否在 Responses API 中启用持久化
-       */
-      store: import_zod6.z.boolean().optional(),
-      /**
-       * 与请求关联的元数据
-       */
-      metadata: import_zod6.z.record(import_zod6.z.string().max(64), import_zod6.z.string().max(512)).optional(),
-      /**
-       * 预测模式的参数
-       */
-      prediction: import_zod6.z.record(import_zod6.z.string(), import_zod6.z.any()).optional(),
-      /**
-       * 请求的服务层级
-       * - 'auto': 默认服务层级
-       * - 'flex': 50% 更便宜但延迟更高（仅限 o3, o4-mini, gpt-5）
-       * - 'priority': 更快处理（需要企业版）
-       * - 'default': 标准定价和性能
-       */
-      serviceTier: import_zod6.z.enum(["auto", "flex", "priority", "default"]).optional(),
-      /**
-       * 是否使用严格的 JSON schema 验证
-       * @default true
-       */
-      strictJsonSchema: import_zod6.z.boolean().optional(),
-      /**
-       * 控制模型响应的详细程度
-       * 较低的值会产生更简洁的响应，较高的值会产生更详细的响应
-       */
-      textVerbosity: import_zod6.z.enum(["low", "medium", "high"]).optional(),
-      /**
-       * 提示缓存的缓存键
-       */
-      promptCacheKey: import_zod6.z.string().optional(),
-      /**
-       * 提示缓存的保留策略
-       * - 'in_memory': 默认，标准提示缓存行为
-       * - '24h': 扩展提示缓存，保持缓存前缀最多 24 小时
-       */
-      promptCacheRetention: import_zod6.z.enum(["in_memory", "24h"]).optional(),
-      /**
-       * 用于帮助检测违反使用政策用户的稳定标识符
-       */
-      safetyIdentifier: import_zod6.z.string().optional(),
-      /**
-       * 覆盖此模型的系统消息模式
-       * - 'system': 使用 'system' 角色（大多数模型的默认值）
-       * - 'developer': 使用 'developer' 角色（推理模型使用）
-       * - 'remove': 完全移除系统消息
-       */
-      systemMessageMode: import_zod6.z.enum(["system", "developer", "remove"]).optional(),
-      /**
-       * 强制将此模型视为推理模型
-       * 适用于自定义 baseURL 的"隐形"推理模型
-       */
-      forceReasoning: import_zod6.z.boolean().optional()
-    })
-  )
-);
-var aihubmixResponsesProviderOptionsSchema = (0, import_provider_utils6.lazySchema)(
-  () => (0, import_provider_utils6.zodSchema)(
-    import_zod6.z.object({
-      /**
-       * OpenAI 对话的 ID，用于继续对话
-       */
-      conversation: import_zod6.z.string().nullish(),
-      /**
-       * 响应中包含的额外字段
-       */
-      include: import_zod6.z.array(
-        import_zod6.z.enum([
-          "reasoning.encrypted_content",
-          "file_search_call.results",
-          "message.output_text.logprobs"
-        ])
-      ).nullish(),
-      /**
-       * 模型的指令
-       */
-      instructions: import_zod6.z.string().nullish(),
-      /**
-       * 返回 token 的对数概率
-       */
-      logprobs: import_zod6.z.union([import_zod6.z.boolean(), import_zod6.z.number().min(1).max(20)]).optional(),
-      /**
-       * 内置工具调用的最大总数
-       */
-      maxToolCalls: import_zod6.z.number().nullish(),
-      /**
-       * 与生成关联的额外元数据
-       */
-      metadata: import_zod6.z.any().nullish(),
-      /**
-       * 是否使用并行工具调用，默认为 true
-       */
-      parallelToolCalls: import_zod6.z.boolean().nullish(),
-      /**
-       * 上一个响应的 ID，用于继续对话
-       */
-      previousResponseId: import_zod6.z.string().nullish(),
-      /**
-       * 提示缓存键
-       */
-      promptCacheKey: import_zod6.z.string().nullish(),
-      /**
-       * 提示缓存保留策略
-       */
-      promptCacheRetention: import_zod6.z.enum(["in_memory", "24h"]).nullish(),
-      /**
-       * 推理模型的推理强度
-       */
-      reasoningEffort: import_zod6.z.string().nullish(),
-      /**
-       * 控制推理摘要输出
-       */
-      reasoningSummary: import_zod6.z.string().nullish(),
-      /**
-       * 安全监控的标识符
-       */
-      safetyIdentifier: import_zod6.z.string().nullish(),
-      /**
-       * 请求的服务层级
-       */
-      serviceTier: import_zod6.z.enum(["auto", "flex", "priority", "default"]).nullish(),
-      /**
-       * 是否存储生成内容，默认为 true
-       */
-      store: import_zod6.z.boolean().nullish(),
-      /**
-       * 是否使用严格的 JSON schema 验证
-       */
-      strictJsonSchema: import_zod6.z.boolean().nullish(),
-      /**
-       * 控制模型响应的详细程度
-       */
-      textVerbosity: import_zod6.z.enum(["low", "medium", "high"]).nullish(),
-      /**
-       * 输出截断控制
-       */
-      truncation: import_zod6.z.enum(["auto", "disabled"]).nullish(),
-      /**
-       * 代表终端用户的唯一标识符
-       */
-      user: import_zod6.z.string().nullish(),
-      /**
-       * 系统消息模式
-       */
-      systemMessageMode: import_zod6.z.enum(["system", "developer", "remove"]).optional(),
-      /**
-       * 强制将模型视为推理模型
-       */
-      forceReasoning: import_zod6.z.boolean().optional()
-    })
-  )
-);
 var AihubmixTranscriptionModel = class extends import_internal.OpenAITranscriptionModel {
   async doGenerate(options) {
     if (options.mediaType) {
@@ -536,29 +349,13 @@ var AihubmixTranscriptionModel = class extends import_internal.OpenAITranscripti
     return super.doGenerate(options);
   }
 };
-var AihubmixOpenAIChatLanguageModel = class _AihubmixOpenAIChatLanguageModel extends import_internal.OpenAIChatLanguageModel {
-  constructor(modelId, settings) {
-    super(modelId, {
-      ...settings,
-      fetch: _AihubmixOpenAIChatLanguageModel.createCustomFetch(settings.fetch)
-    });
+function transformRequestBody(body) {
+  if (body.tools && Array.isArray(body.tools) && body.tools.length === 0 && body.tool_choice) {
+    const { tool_choice, ...rest } = body;
+    return rest;
   }
-  static createCustomFetch(originalFetch) {
-    return async (url, options) => {
-      if (options?.body) {
-        try {
-          const body = JSON.parse(options.body);
-          if (body.tools && Array.isArray(body.tools) && body.tools.length === 0 && body.tool_choice) {
-            delete body.tool_choice;
-            options.body = JSON.stringify(body);
-          }
-        } catch (error) {
-        }
-      }
-      return originalFetch ? originalFetch(url, options) : fetch(url, options);
-    };
-  }
-};
+  return body;
+}
 function createAihubmix(options = {}) {
   const getHeaders = () => ({
     Authorization: `Bearer ${(0, import_provider_utils6.loadApiKey)({
@@ -584,12 +381,13 @@ function createAihubmix(options = {}) {
   const createChatModel = (deploymentName, settings = {}) => {
     const headers = getHeaders();
     if (deploymentName.startsWith("claude-")) {
+      const { Authorization, ...restHeaders } = headers;
       return new import_internal2.AnthropicMessagesLanguageModel(deploymentName, {
         provider: "aihubmix.chat",
         baseURL: url({ path: "", modelId: deploymentName }),
         headers: {
-          ...headers,
-          "x-api-key": headers["Authorization"].split(" ")[1]
+          ...restHeaders,
+          "x-api-key": Authorization.split(" ")[1]
         },
         supportedUrls: () => ({
           "image/*": [/^https?:\/\/.*$/]
@@ -597,59 +395,53 @@ function createAihubmix(options = {}) {
       });
     }
     if ((deploymentName.startsWith("gemini") || deploymentName.startsWith("imagen")) && !deploymentName.endsWith("-nothink") && !deploymentName.endsWith("-search")) {
+      const { Authorization, ...restHeaders } = headers;
       return new import_internal3.GoogleGenerativeAILanguageModel(
         deploymentName,
         {
           provider: "aihubmix.chat",
           baseURL: "https://aihubmix.com/gemini/v1beta",
           headers: {
-            ...headers,
-            "x-goog-api-key": headers["Authorization"].split(" ")[1]
+            ...restHeaders,
+            "x-goog-api-key": Authorization.split(" ")[1]
           },
           generateId: () => `aihubmix-${Date.now()}`,
           supportedUrls: () => ({})
         }
       );
     }
-    if (deploymentName === "gpt-5-pro" || deploymentName === "gpt-5-codex") {
-      return new import_internal.OpenAIResponsesLanguageModel(deploymentName, {
-        provider: "aihubmix.chat",
-        url,
-        headers: getHeaders,
-        fetch: options.fetch,
-        fileIdPrefixes: ["file-"]
-      });
-    }
-    return new AihubmixOpenAIChatLanguageModel(deploymentName, {
+    return new import_openai_compatible.OpenAICompatibleChatLanguageModel(deploymentName, {
       provider: "aihubmix.chat",
       url,
       headers: getHeaders,
-      fetch: options.fetch
+      fetch: options.fetch,
+      includeUsage: true,
+      supportsStructuredOutputs: true,
+      transformRequestBody
     });
   };
-  const createCompletionModel = (modelId, settings = {}) => new import_internal.OpenAICompletionLanguageModel(modelId, {
+  const createCompletionModel = (modelId, settings = {}) => new import_openai_compatible.OpenAICompatibleCompletionLanguageModel(modelId, {
     provider: "aihubmix.completion",
     url,
     headers: getHeaders,
-    fetch: options.fetch
+    fetch: options.fetch,
+    includeUsage: true
   });
   const createEmbeddingModel = (modelId, settings = {}) => {
-    return new import_internal.OpenAIEmbeddingModel(modelId, {
+    return new import_openai_compatible.OpenAICompatibleEmbeddingModel(modelId, {
       provider: "aihubmix.embeddings",
-      headers: getHeaders,
       url,
+      headers: getHeaders,
       fetch: options.fetch
     });
   };
   const createResponsesModel = (modelId) => new import_internal.OpenAIResponsesLanguageModel(modelId, {
     provider: "aihubmix.responses",
     url,
-    headers: getHeaders,
-    fetch: options.fetch,
-    fileIdPrefixes: ["file-"]
+    headers: getHeaders
   });
   const createImageModel = (modelId, settings = {}) => {
-    return new import_internal.OpenAIImageModel(modelId, {
+    return new import_openai_compatible.OpenAICompatibleImageModel(modelId, {
       provider: "aihubmix.image",
       url,
       headers: getHeaders,
@@ -700,8 +492,6 @@ var aihubmix = createAihubmix();
 // Annotate the CommonJS export names for ESM import in node:
 0 && (module.exports = {
   aihubmix,
-  aihubmixChatProviderOptionsSchema,
-  aihubmixResponsesProviderOptionsSchema,
   createAihubmix
 });
 //# sourceMappingURL=index.js.map