@ai-sdk/amazon-bedrock 4.0.86 → 4.0.87

package/docs/08-amazon-bedrock.mdx

@@ -514,6 +514,37 @@ console.log(amazonResult.text); // text response
 See [AI SDK UI: Chatbot](/docs/ai-sdk-ui/chatbot#reasoning) for more details
 on how to integrate reasoning into your chatbot.
 
+## Service Tiers
+
+Amazon Bedrock supports selecting an inference service tier per request via the `serviceTier` provider option.
+
+```ts
+import {
+  bedrock,
+  type AmazonBedrockLanguageModelOptions,
+} from '@ai-sdk/amazon-bedrock';
+import { generateText } from 'ai';
+
+const result = await generateText({
+  model: bedrock('us.anthropic.claude-sonnet-4-20250514-v1:0'),
+  prompt: 'Summarize this support ticket backlog.',
+  providerOptions: {
+    bedrock: {
+      serviceTier: 'priority',
+    } satisfies AmazonBedrockLanguageModelOptions,
+  },
+});
+```
+
+Supported values are:
+
+- `reserved`
+- `priority`
+- `default`
+- `flex`
+
+See the [Amazon Bedrock service tiers documentation](https://docs.aws.amazon.com/bedrock/latest/userguide/service-tiers-inference.html) for model availability and behavior.
+
 ## Extended Context Window
 
 Claude Sonnet 4 models on Amazon Bedrock support an extended context window of up to 1 million tokens when using the `context-1m-2025-08-07` beta feature.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@ai-sdk/amazon-bedrock",
-  "version": "4.0.86",
+  "version": "4.0.87",
   "license": "Apache-2.0",
   "sideEffects": false,
   "main": "./dist/index.js",
@@ -39,8 +39,8 @@
     "@smithy/util-utf8": "^4.0.0",
     "aws4fetch": "^1.0.20",
     "@ai-sdk/anthropic": "3.0.64",
-    "@ai-sdk/provider-utils": "4.0.21",
-    "@ai-sdk/provider": "3.0.8"
+    "@ai-sdk/provider": "3.0.8",
+    "@ai-sdk/provider-utils": "4.0.21"
   },
   "devDependencies": {
     "@types/node": "20.17.24",

package/src/bedrock-api-types.ts

@@ -13,6 +13,9 @@ export interface BedrockConverseInput {
   };
   additionalModelRequestFields?: Record<string, unknown>;
   additionalModelResponseFieldPaths?: string[];
+  serviceTier?: {
+    type: string;
+  };
   guardrailConfig?:
     | BedrockGuardrailConfiguration
     | BedrockGuardrailStreamConfiguration

package/src/bedrock-chat-language-model.ts

@@ -368,6 +368,7 @@ export class BedrockChatLanguageModel implements LanguageModelV3 {
     const {
       reasoningConfig: _,
       additionalModelRequestFields: __,
+      serviceTier: ___,
       ...filteredBedrockOptions
     } = providerOptions?.bedrock || {};
 
@@ -387,6 +388,11 @@ export class BedrockChatLanguageModel implements LanguageModelV3 {
         ...(Object.keys(inferenceConfig).length > 0 && {
           inferenceConfig,
         }),
+        ...(bedrockOptions.serviceTier != null && {
+          serviceTier: {
+            type: bedrockOptions.serviceTier,
+          },
+        }),
         ...filteredBedrockOptions,
         ...(toolConfig.tools !== undefined && toolConfig.tools.length > 0
           ? { toolConfig }

package/src/bedrock-chat-options.ts

@@ -122,6 +122,16 @@ export const amazonBedrockLanguageModelOptions = z.object({
    * Anthropic beta features to enable
    */
   anthropicBeta: z.array(z.string()).optional(),
+  /**
+   * Service tier for the request.
+   * @see https://docs.aws.amazon.com/bedrock/latest/userguide/service-tiers-inference.html
+   *
+   * - 'reserved': Uses provisioned throughput capacity
+   * - 'priority': Prioritizes low-latency inference when capacity is available
+   * - 'default': Standard on-demand tier
+   * - 'flex': Lower-cost tier for flexible latency workloads
+   */
+  serviceTier: z.enum(['reserved', 'priority', 'default', 'flex']).optional(),
 });
 
 export type AmazonBedrockLanguageModelOptions = z.infer<