feat: add UI checkbox for Bedrock backend option in LiteLLM provider

- Add litellmUseAzureBedrock option to provider settings schema
- Add checkbox UI in LiteLLM settings to let users specify Bedrock backend
- Update isBedrockModel() to use user-specified option over auto-detection
- Add comprehensive tests for the new option
- Add English translations for the new setting

packages/types/src/provider-settings.ts

@@ -374,6 +374,7 @@ const litellmSchema = baseProviderSettingsSchema.extend({
 	litellmApiKey: z.string().optional(),
 	litellmModelId: z.string().optional(),
 	litellmUsePromptCache: z.boolean().optional(),
+	litellmUseAzureBedrock: z.boolean().optional(),
 })
 
 const cerebrasSchema = apiModelIdProviderModelSchema.extend({

src/api/providers/__tests__/lite-llm.spec.ts

@@ -401,7 +401,117 @@ describe("LiteLLMHandler", () => {
 	})
 
 	describe("Bedrock model handling", () => {
-		it("should exclude parallel_tool_calls for Bedrock models when using native tools", async () => {
+		it("should exclude parallel_tool_calls when litellmUseAzureBedrock is explicitly true", async () => {
+			const options: ApiHandlerOptions = {
+				...mockOptions,
+				litellmModelId: "gpt-4", // Non-Bedrock model ID
+				litellmUseAzureBedrock: true, // Explicitly set to Bedrock
+			}
+			handler = new LiteLLMHandler(options)
+
+			const systemPrompt = "You are a helpful assistant"
+			const messages: Anthropic.Messages.MessageParam[] = [{ role: "user", content: "Test" }]
+
+			// Mock the stream response
+			const mockStream = {
+				async *[Symbol.asyncIterator]() {
+					yield {
+						choices: [{ delta: { content: "Response" } }],
+						usage: {
+							prompt_tokens: 10,
+							completion_tokens: 5,
+						},
+					}
+				},
+			}
+
+			mockCreate.mockReturnValue({
+				withResponse: vi.fn().mockResolvedValue({ data: mockStream }),
+			})
+
+			const metadata = {
+				taskId: "test-task",
+				tools: [
+					{
+						type: "function" as const,
+						function: {
+							name: "test_tool",
+							description: "A test tool",
+							parameters: { type: "object", properties: {} },
+						},
+					},
+				],
+				toolProtocol: "native" as const,
+				parallelToolCalls: true,
+			}
+
+			const generator = handler.createMessage(systemPrompt, messages, metadata)
+			for await (const chunk of generator) {
+				// Consume the generator
+			}
+
+			// Verify that parallel_tool_calls is NOT included when litellmUseAzureBedrock is true
+			const createCall = mockCreate.mock.calls[0][0]
+			expect(createCall.parallel_tool_calls).toBeUndefined()
+			expect(createCall.tools).toBeDefined()
+		})
+
+		it("should include parallel_tool_calls when litellmUseAzureBedrock is explicitly false", async () => {
+			const options: ApiHandlerOptions = {
+				...mockOptions,
+				litellmModelId: "bedrock/anthropic.claude-3-5-sonnet-20241022-v2:0", // Bedrock model ID
+				litellmUseAzureBedrock: false, // Explicitly set to NOT Bedrock
+			}
+			handler = new LiteLLMHandler(options)
+
+			const systemPrompt = "You are a helpful assistant"
+			const messages: Anthropic.Messages.MessageParam[] = [{ role: "user", content: "Test" }]
+
+			// Mock the stream response
+			const mockStream = {
+				async *[Symbol.asyncIterator]() {
+					yield {
+						choices: [{ delta: { content: "Response" } }],
+						usage: {
+							prompt_tokens: 10,
+							completion_tokens: 5,
+						},
+					}
+				},
+			}
+
+			mockCreate.mockReturnValue({
+				withResponse: vi.fn().mockResolvedValue({ data: mockStream }),
+			})
+
+			const metadata = {
+				taskId: "test-task",
+				tools: [
+					{
+						type: "function" as const,
+						function: {
+							name: "test_tool",
+							description: "A test tool",
+							parameters: { type: "object", properties: {} },
+						},
+					},
+				],
+				toolProtocol: "native" as const,
+				parallelToolCalls: true,
+			}
+
+			const generator = handler.createMessage(systemPrompt, messages, metadata)
+			for await (const chunk of generator) {
+				// Consume the generator
+			}
+
+			// Verify that parallel_tool_calls IS included when litellmUseAzureBedrock is false
+			const createCall = mockCreate.mock.calls[0][0]
+			expect(createCall.parallel_tool_calls).toBe(true)
+			expect(createCall.tools).toBeDefined()
+		})
+
+		it("should auto-detect and exclude parallel_tool_calls for Bedrock models when litellmUseAzureBedrock is not set", async () => {
 			const bedrockModels = ["bedrock/anthropic.claude-3-5-sonnet-20241022-v2:0", "amazon.titan-text-express-v1"]
 
 			for (const modelId of bedrockModels) {
@@ -461,7 +571,7 @@ describe("LiteLLMHandler", () => {
 			}
 		})
 
-		it("should include parallel_tool_calls for non-Bedrock models when using native tools", async () => {
+		it("should auto-detect and include parallel_tool_calls for non-Bedrock models when litellmUseAzureBedrock is not set", async () => {
 			const nonBedrockModels = [
 				"gpt-4",
 				"claude-3-opus",

src/api/providers/lite-llm.ts

@@ -41,9 +41,18 @@ export class LiteLLMHandler extends RouterProvider implements SingleCompletionHa
 	/**
 	 * Check if the model is routed through AWS Bedrock
 	 * Bedrock doesn't support the parallel_tool_calls parameter
+	 *
+	 * If the user has explicitly set litellmUseAzureBedrock, use that setting.
+	 * Otherwise, fall back to auto-detection based on model ID patterns.
 	 * Note: We exclude 'anthropic.' prefix as it can match direct Anthropic API access through LiteLLM
 	 */
 	private isBedrockModel(modelId: string): boolean {
+		// User-specified option takes precedence
+		if (this.options.litellmUseAzureBedrock !== undefined) {
+			return this.options.litellmUseAzureBedrock
+		}
+
+		// Fall back to auto-detection
 		const lowerModel = modelId.toLowerCase()
 		return (
 			lowerModel.includes("bedrock") ||

webview-ui/src/components/settings/providers/LiteLLM.tsx

@@ -156,6 +156,20 @@ export const LiteLLM = ({
 				simplifySettings={simplifySettings}
 			/>
 
+			{/* Bedrock backend option */}
+			<div className="mt-4">
+				<VSCodeCheckbox
+					checked={apiConfiguration.litellmUseAzureBedrock || false}
+					onChange={(e: any) => {
+						setApiConfigurationField("litellmUseAzureBedrock", e.target.checked)
+					}}>
+					<span className="font-medium">{t("settings:providers.litellmUseAzureBedrock")}</span>
+				</VSCodeCheckbox>
+				<div className="text-sm text-vscode-descriptionForeground ml-6 mt-1">
+					{t("settings:providers.litellmUseAzureBedrockDescription")}
+				</div>
+			</div>
+
 			{/* Show prompt caching option if the selected model supports it */}
 			{(() => {
 				const selectedModelId = apiConfiguration.litellmModelId || litellmDefaultModelId

webview-ui/src/i18n/locales/en/settings.json

@@ -370,6 +370,8 @@
 		"getXaiApiKey": "Get xAI API Key",
 		"litellmApiKey": "LiteLLM API Key",
 		"litellmBaseUrl": "LiteLLM Base URL",
+		"litellmUseAzureBedrock": "Backend is AWS Bedrock",
+		"litellmUseAzureBedrockDescription": "Enable this if your LiteLLM proxy routes to AWS Bedrock models. This ensures Bedrock-incompatible parameters are excluded from requests.",
 		"awsCredentials": "AWS Credentials",
 		"awsProfile": "AWS Profile",
 		"awsApiKey": "Amazon Bedrock API Key",