From db95bdedb5f69f829957dad1cfaf30f579a27374 Mon Sep 17 00:00:00 2001
From: koureasstavros <koureasstavros@gmail.com>
Date: Mon, 3 Nov 2025 12:34:21 +0200
Subject: [PATCH 01/21] Add support for Azure Foundry models using responses
 api

---
 .../byok/vscode-node/azureProvider.ts         | 180 ++++++++++++++++--
 .../byok/vscode-node/customOAIProvider.ts     |   4 +-
 .../codeBlocks/node/codeBlockProcessor.ts     |  39 ++--
 .../common/configurationService.ts            |   2 +-
 4 files changed, 186 insertions(+), 39 deletions(-)

diff --git a/src/extension/byok/vscode-node/azureProvider.ts b/src/extension/byok/vscode-node/azureProvider.ts
index 673e8ca845..e0a7db025e 100644
--- a/src/extension/byok/vscode-node/azureProvider.ts
+++ b/src/extension/byok/vscode-node/azureProvider.ts
@@ -4,15 +4,27 @@
  *--------------------------------------------------------------------------------------------*/
 
 import { ConfigKey, IConfigurationService } from '../../../platform/configuration/common/configurationService';
+import { IChatModelInformation, ModelSupportedEndpoint } from '../../../platform/endpoint/common/endpointProvider';
 import { ILogService } from '../../../platform/log/common/logService';
 import { IExperimentationService } from '../../../platform/telemetry/common/nullExperimentationService';
 import { IInstantiationService } from '../../../util/vs/platform/instantiation/common/instantiation';
+import { BYOKModelCapabilities } from '../common/byokProvider';
 import { IBYOKStorageService } from './byokStorageService';
 import { CustomOAIBYOKModelProvider, hasExplicitApiPath } from './customOAIProvider';
 
-export function resolveAzureUrl(modelId: string, url: string): string {
+export interface AzureUrlOptions {
+	deploymentType?: 'completions' | 'responses';
+	deploymentName?: string;
+	apiVersion?: string;
+}
+
+export function resolveAzureUrl(modelId: string, url: string, options?: AzureUrlOptions, logService?: ILogService): string {
+	logService?.info(`[Azure Debug] resolveAzureUrl called - modelId: ${modelId}, url: ${url}`);
+	logService?.info(`[Azure Debug] options: ${JSON.stringify(options, null, 2)}`);
+
 	// The fully resolved url was already passed in
 	if (hasExplicitApiPath(url)) {
+		logService?.info(`[Azure Debug] URL has explicit API path, returning as-is: ${url}`);
 		return url;
 	}
 
@@ -25,15 +37,50 @@ export function resolveAzureUrl(modelId: string, url: string): string {
 		url = url.slice(0, -3);
 	}
 
-	// Default to chat completions for base URLs
-	const defaultApiPath = '/chat/completions';
+	// Check if deployment type is specified
+	// If no deployment name is provided, use modelId as fallback
+	const deploymentName = options?.deploymentName || modelId;
+	const deploymentType = options?.deploymentType || 'completions';
+	const apiVersion = options?.apiVersion || '2025-01-01-preview';
+
+	logService?.info(`[Azure Debug] Resolved values - deploymentName: ${deploymentName}, deploymentType: ${deploymentType}, apiVersion: ${apiVersion}`);
+
+	// Determine if this is an Azure OpenAI endpoint (requires deployment name in path)
+	const isAzureOpenAIEndpointv1 = url.includes('models.ai.azure.com') || url.includes('inference.ml.azure.com');
+	const isAzureOpenAIEndpoint = url.includes('openai.azure.com') || url.includes('cognitiveservices.azure.com');
+
+	logService?.info(`[Azure Debug] Endpoint detection - v1: ${isAzureOpenAIEndpointv1}, standard: ${isAzureOpenAIEndpoint}`);
 
-	if (url.includes('models.ai.azure.com') || url.includes('inference.ml.azure.com')) {
-		return `${url}/v1${defaultApiPath}`;
-	} else if (url.includes('openai.azure.com')) {
-		return `${url}/openai/deployments/${modelId}${defaultApiPath}?api-version=2025-01-01-preview`;
+	if (deploymentType === 'responses') {
+		// Handle Responses API
+		// Deployment name is passed in the request body as 'model' parameter, not in URL
+		let resolvedUrl: string;
+		if (isAzureOpenAIEndpointv1) {
+			resolvedUrl = `${url}/v1/responses?api-version=${apiVersion}`;
+		} else if (isAzureOpenAIEndpoint) {
+			resolvedUrl = `${url}/openai/responses?api-version=${apiVersion}`;
+		} else {
+			throw new Error(`Unrecognized Azure deployment URL for Responses API: ${url}`);
+		}
+		logService?.info(`[Azure Debug] Responses API URL resolved to: ${resolvedUrl}`);
+		logService?.info(`[Azure Debug] NOTE: Deployment name '${deploymentName}' will be sent in request body as 'model' field`);
+		return resolvedUrl;
+	} else if (deploymentType === 'completions') {
+		// Handle Chat Completions API (default)
+		const defaultApiPath = '/chat/completions';
+		let resolvedUrl: string;
+
+		if (isAzureOpenAIEndpointv1) {
+			resolvedUrl = `${url}/v1${defaultApiPath}`;
+		} else if (isAzureOpenAIEndpoint) {
+			resolvedUrl = `${url}/openai/deployments/${deploymentName}${defaultApiPath}?api-version=${apiVersion}`;
+		} else {
+			throw new Error(`Unrecognized Azure deployment URL: ${url}`);
+		}
+		logService?.info(`[Azure Debug] Completions API URL resolved to: ${resolvedUrl}`);
+		return resolvedUrl;
 	} else {
-		throw new Error(`Unrecognized Azure deployment URL: ${url}`);
+		throw new Error(`Invalid deployment type specified for model ${modelId}: ${deploymentType}`);
 	}
 }
 
@@ -41,21 +88,26 @@ export class AzureBYOKModelProvider extends CustomOAIBYOKModelProvider {
 	static override readonly providerName = 'Azure';
 
 	constructor(
-		byokStorageService: IBYOKStorageService,
-		@IConfigurationService configurationService: IConfigurationService,
-		@ILogService logService: ILogService,
-		@IInstantiationService instantiationService: IInstantiationService,
-		@IExperimentationService experimentationService: IExperimentationService
+		_byokStorageService: IBYOKStorageService,
+		@IConfigurationService _configurationService: IConfigurationService,
+		@ILogService _logService: ILogService,
+		@IInstantiationService _instantiationService: IInstantiationService,
+		@IExperimentationService _experimentationService: IExperimentationService
 	) {
 		super(
-			byokStorageService,
-			configurationService,
-			logService,
-			instantiationService,
-			experimentationService
+			_byokStorageService,
+			_configurationService,
+			_logService,
+			_instantiationService,
+			_experimentationService
 		);
 		// Override the instance properties
 		this.providerName = AzureBYOKModelProvider.providerName;
+
+		// Bind methods to ensure proper context
+		this.resolveUrl = this.resolveUrl.bind(this);
+		this.getModelInfo = this.getModelInfo.bind(this);
+		this.getConfigKey = this.getConfigKey.bind(this);
 	}
 
 	protected override getConfigKey() {
@@ -63,6 +115,96 @@ export class AzureBYOKModelProvider extends CustomOAIBYOKModelProvider {
 	}
 
 	protected override resolveUrl(modelId: string, url: string): string {
-		return resolveAzureUrl(modelId, url);
+		this._logService?.info(`[Azure Debug] resolveUrl called for modelId: ${modelId}`);
+		try {
+			// Get model config to access deployment options
+			const modelConfig = this._configurationService?.getConfig(this.getConfigKey()) as Record<string, {
+				deploymentType?: 'completions' | 'responses';
+				deploymentName?: string;
+				apiVersion?: string;
+				url: string;
+			}> | undefined;
+
+			const config = modelConfig?.[modelId];
+			this._logService?.info(`[Azure Debug] Model config for ${modelId}: ${JSON.stringify(config, null, 2)}`);
+
+			const options: AzureUrlOptions | undefined = config ? {
+				deploymentType: config.deploymentType,
+				deploymentName: config.deploymentName,
+				apiVersion: config.apiVersion
+			} : undefined;
+
+			const resolvedUrl = resolveAzureUrl(modelId, url, options, this._logService);
+			this._logService?.info(`[Azure Debug] Final resolved URL: ${resolvedUrl}`);
+			return resolvedUrl;
+		} catch (error) {
+			this._logService?.error(`AzureBYOKModelProvider: Error resolving URL for model ${modelId}:`, error);
+			// Fallback to basic URL resolution
+			return resolveAzureUrl(modelId, url, undefined, this._logService);
+		}
+	}
+
+	protected override async getModelInfo(modelId: string, apiKey: string | undefined, modelCapabilities?: BYOKModelCapabilities): Promise<IChatModelInformation> {
+		this._logService?.info(`[Azure Debug] getModelInfo called for modelId: ${modelId}`);
+		this._logService?.info(`[Azure Debug] Initial modelCapabilities: ${JSON.stringify(modelCapabilities, null, 2)}`);
+		try {
+			// Get model config to check deployment type and deployment name
+			const configKey = this.getConfigKey();
+			this._logService?.info(`[Azure Debug] Config key: ${configKey}`);
+
+			const modelConfig = this._configurationService?.getConfig(configKey);
+			this._logService?.info(`[Azure Debug] Raw modelConfig type: ${typeof modelConfig}, value: ${JSON.stringify(modelConfig, null, 2)}`);
+
+			// Safely access the model-specific config
+			let config: { deploymentType?: 'completions' | 'responses'; deploymentName?: string } | undefined;
+			if (modelConfig && typeof modelConfig === 'object' && modelId in modelConfig) {
+				config = (modelConfig as Record<string, any>)[modelId];
+			}
+
+			this._logService?.info(`[Azure Debug] Model config for ${modelId}: ${JSON.stringify(config, null, 2)}`);
+
+			const deploymentType = config?.deploymentType || 'completions';
+			// If no deployment name is provided, use modelId as fallback
+			const deploymentName = config?.deploymentName || modelId;
+
+			this._logService?.info(`[Azure Debug] Deployment info - type: ${deploymentType}, name: ${deploymentName}`);
+
+			// For Azure Responses API, the 'model' field in the request body must be the deployment name
+			// We override the modelCapabilities.name to be the deployment name so it gets used correctly
+			let updatedCapabilities = modelCapabilities;
+			if (deploymentType === 'responses' && modelCapabilities) {
+				updatedCapabilities = {
+					...modelCapabilities,
+					name: deploymentName
+				};
+				this._logService?.info(`[Azure Debug] RESPONSES API: Overriding model name from '${modelCapabilities.name}' to '${deploymentName}'`);
+				this._logService?.info(`[Azure Debug] Updated capabilities: ${JSON.stringify(updatedCapabilities, null, 2)}`);
+			} else {
+				this._logService?.info(`[Azure Debug] COMPLETIONS API: Using capabilities as-is (deployment name in URL)`);
+			}
+
+			const modelInfo = await super.getModelInfo(modelId, apiKey, updatedCapabilities);
+
+			// Always set modelInfo.id to deployment name for Azure deployments
+			modelInfo.id = deploymentName;
+
+			// Set supported endpoints based on deployment type
+			if (deploymentType === 'responses') {
+				modelInfo.supported_endpoints = [ModelSupportedEndpoint.Responses];
+				this._logService?.info(`[Azure Debug] Set supported_endpoints to: [Responses]`);
+			} else {
+				// For completions API, only support chat completions
+				modelInfo.supported_endpoints = [ModelSupportedEndpoint.ChatCompletions];
+				this._logService?.info(`[Azure Debug] Set supported_endpoints to: [ChatCompletions]`);
+			}
+			this._logService?.info(`[Azure Debug] Set modelInfo.id to deployment name: '${deploymentName}'`);
+			this._logService?.info(`[Azure Debug] Final modelInfo.id: ${modelInfo.id}`);
+			this._logService?.info(`[Azure Debug] Final modelInfo.supported_endpoints: ${JSON.stringify(modelInfo.supported_endpoints)}`);
+
+			return modelInfo;
+		} catch (error) {
+			this._logService?.error(`AzureBYOKModelProvider: Error getting model info for ${modelId}:`, error);
+			throw error;
+		}
 	}
 }
diff --git a/src/extension/byok/vscode-node/customOAIProvider.ts b/src/extension/byok/vscode-node/customOAIProvider.ts
index 825226dd4c..bc75d0fc6d 100644
--- a/src/extension/byok/vscode-node/customOAIProvider.ts
+++ b/src/extension/byok/vscode-node/customOAIProvider.ts
@@ -86,8 +86,8 @@ export class CustomOAIBYOKModelProvider implements BYOKModelProvider<CustomOAIMo
 		return modelInfo;
 	}
 
-	private getUserModelConfig(): Record<string, { name: string; url: string; toolCalling: boolean; vision: boolean; maxInputTokens: number; maxOutputTokens: number; requiresAPIKey: boolean; thinking?: boolean; editTools?: EndpointEditToolName[]; requestHeaders?: Record<string, string> }> {
-		const modelConfig = this._configurationService.getConfig(this.getConfigKey()) as Record<string, { name: string; url: string; toolCalling: boolean; vision: boolean; maxInputTokens: number; maxOutputTokens: number; requiresAPIKey: boolean; thinking?: boolean; editTools?: EndpointEditToolName[]; requestHeaders?: Record<string, string> }>;
+	private getUserModelConfig(): Record<string, { name: string; url: string; toolCalling: boolean; vision: boolean; maxInputTokens: number; maxOutputTokens: number; requiresAPIKey: boolean; thinking?: boolean; editTools?: EndpointEditToolName[]; requestHeaders?: Record<string, string>; deploymentType?: 'completions' | 'responses'; deploymentName?: string; apiVersion?: string }> {
+		const modelConfig = this._configurationService.getConfig(this.getConfigKey()) as Record<string, { name: string; url: string; toolCalling: boolean; vision: boolean; maxInputTokens: number; maxOutputTokens: number; requiresAPIKey: boolean; thinking?: boolean; editTools?: EndpointEditToolName[]; requestHeaders?: Record<string, string>; deploymentType?: 'completions' | 'responses'; deploymentName?: string; apiVersion?: string }>;
 		return modelConfig;
 	}
 
diff --git a/src/extension/codeBlocks/node/codeBlockProcessor.ts b/src/extension/codeBlocks/node/codeBlockProcessor.ts
index a7788366ae..c5671f5fd8 100644
--- a/src/extension/codeBlocks/node/codeBlockProcessor.ts
+++ b/src/extension/codeBlocks/node/codeBlockProcessor.ts
@@ -102,26 +102,31 @@ export class CodeBlockTrackingChatResponseStream implements ChatResponseStream {
 		return new CodeBlocksMetadata(this._codeBlocks);
 	}
 
-	private forward(fc: CallableFunction) {
-		return (...args: any[]) => {
+	private forward<K extends keyof ChatResponseStream>(methodName: K): ChatResponseStream[K] {
+		const method = this._wrapped[methodName];
+		if (typeof method !== 'function') {
+			// Return a no-op function for missing methods
+			return ((() => { }) as any);
+		}
+		return ((...args: any[]) => {
 			this._codeBlockProcessor.flush();
-			return fc(...args);
-		};
+			return (method as Function).apply(this._wrapped, args);
+		}) as any;
 	}
 
-	button = this.forward(this._wrapped.button.bind(this._wrapped));
-	filetree = this.forward(this._wrapped.filetree.bind(this._wrapped));
-	progress = this._wrapped.progress.bind(this._wrapped);
-	reference = this.forward(this._wrapped.reference.bind(this._wrapped));
-	textEdit = this.forward(this._wrapped.textEdit.bind(this._wrapped));
-	notebookEdit = this.forward(this._wrapped.notebookEdit.bind(this._wrapped));
-	confirmation = this.forward(this._wrapped.confirmation.bind(this._wrapped));
-	warning = this.forward(this._wrapped.warning.bind(this._wrapped));
-	reference2 = this.forward(this._wrapped.reference2.bind(this._wrapped));
-	codeCitation = this.forward(this._wrapped.codeCitation.bind(this._wrapped));
-	anchor = this.forward(this._wrapped.anchor.bind(this._wrapped));
-	externalEdit = this.forward(this._wrapped.externalEdit.bind(this._wrapped));
-	prepareToolInvocation = this.forward(this._wrapped.prepareToolInvocation.bind(this._wrapped));
+	button = this.forward('button');
+	filetree = this.forward('filetree');
+	progress = this.forward('progress');
+	reference = this.forward('reference');
+	textEdit = this.forward('textEdit');
+	notebookEdit = this.forward('notebookEdit');
+	confirmation = this.forward('confirmation');
+	warning = this.forward('warning');
+	reference2 = this.forward('reference2');
+	codeCitation = this.forward('codeCitation');
+	anchor = this.forward('anchor');
+	externalEdit = this.forward('externalEdit');
+	prepareToolInvocation = this.forward('prepareToolInvocation');
 }
 
 
diff --git a/src/platform/configuration/common/configurationService.ts b/src/platform/configuration/common/configurationService.ts
index c45571f14b..2ca0159139 100644
--- a/src/platform/configuration/common/configurationService.ts
+++ b/src/platform/configuration/common/configurationService.ts
@@ -811,7 +811,7 @@ export namespace ConfigKey {
 	export const CurrentEditorAgentContext = defineSetting<boolean>('chat.agent.currentEditorContext.enabled', true);
 	/** BYOK  */
 	export const OllamaEndpoint = defineSetting<string>('chat.byok.ollamaEndpoint', 'http://localhost:11434');
-	export const AzureModels = defineSetting<Record<string, { name: string; url: string; toolCalling: boolean; vision: boolean; maxInputTokens: number; maxOutputTokens: number; requiresAPIKey?: boolean; thinking?: boolean }>>('chat.azureModels', {});
+	export const AzureModels = defineSetting<Record<string, { name: string; url: string; deploymentType: 'completions' | 'responses'; deploymentName: string; apiVersion: string; toolCalling: boolean; vision: boolean; maxInputTokens: number; maxOutputTokens: number; requiresAPIKey?: boolean; thinking?: boolean }>>('chat.azureModels', {});
 	export const CustomOAIModels = defineSetting<Record<string, { name: string; url: string; toolCalling: boolean; vision: boolean; maxInputTokens: number; maxOutputTokens: number; requiresAPIKey?: boolean; thinking?: boolean; requestHeaders?: Record<string, string> }>>('chat.customOAIModels', {});
 	export const AutoFixDiagnostics = defineExpSetting<boolean>('chat.agent.autoFix', true);
 	export const NotebookFollowCellExecution = defineSetting<boolean>('chat.notebook.followCellExecution.enabled', false);

From e59cf3cbabed415612042747acfe5bc2e933b000 Mon Sep 17 00:00:00 2001
From: koureasstavros <koureasstavros@gmail.com>
Date: Mon, 3 Nov 2025 12:56:19 +0200
Subject: [PATCH 02/21] Add support for Azure Foundry models using temperature

---
 package.json                                  | 28 +++++++++++++++----
 src/extension/byok/common/byokProvider.ts     |  4 +++
 src/extension/byok/node/openAIEndpoint.ts     |  6 ++++
 .../byok/vscode-node/azureProvider.ts         |  9 +++++-
 .../byok/vscode-node/customOAIProvider.ts     | 13 ++++++---
 .../endpoint/common/endpointProvider.ts       |  1 +
 6 files changed, 50 insertions(+), 11 deletions(-)

diff --git a/package.json b/package.json
index 4f4a722f27..d2be7301cf 100644
--- a/package.json
+++ b/package.json
@@ -3035,13 +3035,17 @@
 									"type": "string",
 									"description": "URL endpoint for the Azure model"
 								},
-								"toolCalling": {
-									"type": "boolean",
-									"description": "Whether the model supports tool calling"
+								"deploymentName": {
+									"type": "string",
+									"description": "Azure deployment name for the model"
 								},
-								"vision": {
-									"type": "boolean",
-									"description": "Whether the model supports vision capabilities"
+								"deploymentType": {
+									"type": "string",
+									"description": "Type of Azure deployment"
+								},
+								"apiVersion": {
+									"type": "string",
+									"description": "Azure OpenAI API version to use"
 								},
 								"maxInputTokens": {
 									"type": "number",
@@ -3051,11 +3055,23 @@
 									"type": "number",
 									"description": "Maximum number of output tokens supported by the model"
 								},
+								"toolCalling": {
+									"type": "boolean",
+									"description": "Whether the model supports tool calling"
+								},
+								"vision": {
+									"type": "boolean",
+									"description": "Whether the model supports vision capabilities"
+								},
 								"thinking": {
 									"type": "boolean",
 									"default": false,
 									"description": "Whether the model supports thinking capabilities"
 								},
+								"temperature": {
+									"type": "number",
+									"description": "Temperature parameter for model responses (typically 0.0 to 2.0)"
+								},
 								"requestHeaders": {
 									"type": "object",
 									"description": "Additional HTTP headers to include with requests to this model. These reserved headers are not allowed and ignored if present: forbidden request headers (https://developer.mozilla.org/en-US/docs/Glossary/Forbidden_request_header), forwarding headers ('forwarded', 'x-forwarded-for', 'x-forwarded-host', 'x-forwarded-proto'), and others ('api-key', 'authorization', 'content-type', 'openai-intent', 'x-github-api-version', 'x-initiator', 'x-interaction-id', 'x-interaction-type', 'x-onbehalf-extension-id', 'x-request-id', 'x-vscode-user-agent-library-version'). Pattern-based forbidden headers ('proxy-*', 'sec-*', 'x-http-method*' with forbidden methods) are also blocked.",
diff --git a/src/extension/byok/common/byokProvider.ts b/src/extension/byok/common/byokProvider.ts
index b4c1bd4928..8d23d3c1f1 100644
--- a/src/extension/byok/common/byokProvider.ts
+++ b/src/extension/byok/common/byokProvider.ts
@@ -55,6 +55,7 @@ export interface BYOKModelCapabilities {
 	toolCalling: boolean;
 	vision: boolean;
 	thinking?: boolean;
+	temperature?: number;
 	editTools?: EndpointEditToolName[];
 	requestHeaders?: Record<string, string>;
 }
@@ -153,6 +154,9 @@ export function resolveModelInfo(modelId: string, providerName: string, knownMod
 	if (knownModelInfo?.requestHeaders && Object.keys(knownModelInfo.requestHeaders).length > 0) {
 		modelInfo.requestHeaders = { ...knownModelInfo.requestHeaders };
 	}
+	if (knownModelInfo?.temperature !== undefined) {
+		modelInfo.temperature = knownModelInfo.temperature;
+	}
 	return modelInfo;
 }
 
diff --git a/src/extension/byok/node/openAIEndpoint.ts b/src/extension/byok/node/openAIEndpoint.ts
index 63fd236ccb..a8e0182c73 100644
--- a/src/extension/byok/node/openAIEndpoint.ts
+++ b/src/extension/byok/node/openAIEndpoint.ts
@@ -113,6 +113,7 @@ export class OpenAIEndpoint extends ChatEndpoint {
 	private static readonly _maxCustomHeaderCount = 20;
 
 	private readonly _customHeaders: Record<string, string>;
+	private readonly _temperature: number | undefined;
 	constructor(
 		_modelMetadata: IChatModelInformation,
 		protected readonly _apiKey: string,
@@ -144,6 +145,7 @@ export class OpenAIEndpoint extends ChatEndpoint {
 			logService
 		);
 		this._customHeaders = this._sanitizeCustomHeaders(_modelMetadata.requestHeaders);
+		this._temperature = _modelMetadata.temperature;
 	}
 
 	private _sanitizeCustomHeaders(headers: Readonly<Record<string, string>> | undefined): Record<string, string> {
@@ -294,6 +296,10 @@ export class OpenAIEndpoint extends ChatEndpoint {
 				body['max_completion_tokens'] = body.max_tokens;
 				delete body.max_tokens;
 			}
+			// Apply temperature from config if specified (and not a thinking model)
+			if (this._temperature !== undefined && !this.modelMetadata.capabilities.supports.thinking) {
+				body.temperature = this._temperature;
+			}
 			// Removing max tokens defaults to the maximum which is what we want for BYOK
 			delete body.max_tokens;
 			if (!this.useResponsesApi && body.stream) {
diff --git a/src/extension/byok/vscode-node/azureProvider.ts b/src/extension/byok/vscode-node/azureProvider.ts
index e0a7db025e..439b89961a 100644
--- a/src/extension/byok/vscode-node/azureProvider.ts
+++ b/src/extension/byok/vscode-node/azureProvider.ts
@@ -122,6 +122,7 @@ export class AzureBYOKModelProvider extends CustomOAIBYOKModelProvider {
 				deploymentType?: 'completions' | 'responses';
 				deploymentName?: string;
 				apiVersion?: string;
+				temperature?: number;
 				url: string;
 			}> | undefined;
 
@@ -156,7 +157,7 @@ export class AzureBYOKModelProvider extends CustomOAIBYOKModelProvider {
 			this._logService?.info(`[Azure Debug] Raw modelConfig type: ${typeof modelConfig}, value: ${JSON.stringify(modelConfig, null, 2)}`);
 
 			// Safely access the model-specific config
-			let config: { deploymentType?: 'completions' | 'responses'; deploymentName?: string } | undefined;
+			let config: { deploymentType?: 'completions' | 'responses'; deploymentName?: string; temperature?: number } | undefined;
 			if (modelConfig && typeof modelConfig === 'object' && modelId in modelConfig) {
 				config = (modelConfig as Record<string, any>)[modelId];
 			}
@@ -188,6 +189,12 @@ export class AzureBYOKModelProvider extends CustomOAIBYOKModelProvider {
 			// Always set modelInfo.id to deployment name for Azure deployments
 			modelInfo.id = deploymentName;
 
+			// Set temperature from config if specified
+			if (config?.temperature !== undefined) {
+				modelInfo.temperature = config.temperature;
+				this._logService?.info(`[Azure Debug] Set temperature from config: ${config.temperature}`);
+			}
+
 			// Set supported endpoints based on deployment type
 			if (deploymentType === 'responses') {
 				modelInfo.supported_endpoints = [ModelSupportedEndpoint.Responses];
diff --git a/src/extension/byok/vscode-node/customOAIProvider.ts b/src/extension/byok/vscode-node/customOAIProvider.ts
index bc75d0fc6d..bdc2a36a37 100644
--- a/src/extension/byok/vscode-node/customOAIProvider.ts
+++ b/src/extension/byok/vscode-node/customOAIProvider.ts
@@ -48,6 +48,7 @@ interface CustomOAIModelInfo extends LanguageModelChatInformation {
 	url: string;
 	thinking: boolean;
 	requestHeaders?: Record<string, string>;
+	temperature?: number;
 }
 
 export class CustomOAIBYOKModelProvider implements BYOKModelProvider<CustomOAIModelInfo> {
@@ -86,8 +87,8 @@ export class CustomOAIBYOKModelProvider implements BYOKModelProvider<CustomOAIMo
 		return modelInfo;
 	}
 
-	private getUserModelConfig(): Record<string, { name: string; url: string; toolCalling: boolean; vision: boolean; maxInputTokens: number; maxOutputTokens: number; requiresAPIKey: boolean; thinking?: boolean; editTools?: EndpointEditToolName[]; requestHeaders?: Record<string, string>; deploymentType?: 'completions' | 'responses'; deploymentName?: string; apiVersion?: string }> {
-		const modelConfig = this._configurationService.getConfig(this.getConfigKey()) as Record<string, { name: string; url: string; toolCalling: boolean; vision: boolean; maxInputTokens: number; maxOutputTokens: number; requiresAPIKey: boolean; thinking?: boolean; editTools?: EndpointEditToolName[]; requestHeaders?: Record<string, string>; deploymentType?: 'completions' | 'responses'; deploymentName?: string; apiVersion?: string }>;
+	private getUserModelConfig(): Record<string, { name: string; url: string; toolCalling: boolean; vision: boolean; maxInputTokens: number; maxOutputTokens: number; requiresAPIKey: boolean; thinking?: boolean; editTools?: EndpointEditToolName[]; requestHeaders?: Record<string, string>; deploymentType?: 'completions' | 'responses'; deploymentName?: string; apiVersion?: string; temperature?: number }> {
+		const modelConfig = this._configurationService.getConfig(this.getConfigKey()) as Record<string, { name: string; url: string; toolCalling: boolean; vision: boolean; maxInputTokens: number; maxOutputTokens: number; requiresAPIKey: boolean; thinking?: boolean; editTools?: EndpointEditToolName[]; requestHeaders?: Record<string, string>; deploymentType?: 'completions' | 'responses'; deploymentName?: string; apiVersion?: string; temperature?: number }>;
 		return modelConfig;
 	}
 
@@ -113,7 +114,8 @@ export class CustomOAIBYOKModelProvider implements BYOKModelProvider<CustomOAIMo
 				maxOutputTokens: modelInfo.maxOutputTokens,
 				thinking: modelInfo.thinking,
 				editTools: modelInfo.editTools,
-				requestHeaders: modelInfo.requestHeaders ? { ...modelInfo.requestHeaders } : undefined
+				requestHeaders: modelInfo.requestHeaders ? { ...modelInfo.requestHeaders } : undefined,
+				temperature: modelInfo.temperature
 			};
 		}
 		return models;
@@ -160,6 +162,7 @@ export class CustomOAIBYOKModelProvider implements BYOKModelProvider<CustomOAIMo
 			},
 			thinking: capabilities.thinking || false,
 			requestHeaders: capabilities.requestHeaders,
+			temperature: capabilities.temperature,
 		};
 		return baseInfo;
 	}
@@ -199,6 +202,7 @@ export class CustomOAIBYOKModelProvider implements BYOKModelProvider<CustomOAIMo
 			thinking: model.thinking,
 			editTools: model.capabilities.editTools?.filter(isEndpointEditToolName),
 			requestHeaders: model.requestHeaders,
+			temperature: model.temperature,
 		});
 		const openAIChatEndpoint = this._instantiationService.createInstance(OpenAIEndpoint, modelInfo, apiKey ?? '', model.url);
 		return this._lmWrapper.provideLanguageModelResponse(openAIChatEndpoint, messages, options, options.requestInitiator, progress, token);
@@ -223,7 +227,8 @@ export class CustomOAIBYOKModelProvider implements BYOKModelProvider<CustomOAIMo
 			name: model.name,
 			url: model.url,
 			thinking: model.thinking,
-			requestHeaders: model.requestHeaders
+			requestHeaders: model.requestHeaders,
+			temperature: model.temperature,
 		});
 		const openAIChatEndpoint = this._instantiationService.createInstance(OpenAIEndpoint, modelInfo, apiKey ?? '', model.url);
 		return this._lmWrapper.provideTokenCount(openAIChatEndpoint, text);
diff --git a/src/platform/endpoint/common/endpointProvider.ts b/src/platform/endpoint/common/endpointProvider.ts
index d60b3c9c78..4e52fd38d6 100644
--- a/src/platform/endpoint/common/endpointProvider.ts
+++ b/src/platform/endpoint/common/endpointProvider.ts
@@ -93,6 +93,7 @@ export type IChatModelInformation = IModelAPIResponse & {
 	capabilities: IChatModelCapabilities;
 	urlOrRequestMetadata?: string | RequestMetadata;
 	requestHeaders?: Readonly<Record<string, string>>;
+	temperature?: number;
 };
 
 export function isChatModelInformation(model: IModelAPIResponse): model is IChatModelInformation {

From cab7654d7bfb97df1b59cc32ce97e41599f923ff Mon Sep 17 00:00:00 2001
From: koureasstavros <koureasstavros@gmail.com>
Date: Mon, 3 Nov 2025 12:59:25 +0200
Subject: [PATCH 03/21] Add settings instructions for Azure Foundry models
 improvements

---
 .../azure-provider-settings.instructions.md   | 273 ++++++++++++++++++
 1 file changed, 273 insertions(+)
 create mode 100644 .github/instructions/azure-provider-settings.instructions.md

diff --git a/.github/instructions/azure-provider-settings.instructions.md b/.github/instructions/azure-provider-settings.instructions.md
new file mode 100644
index 0000000000..afd840a021
--- /dev/null
+++ b/.github/instructions/azure-provider-settings.instructions.md
@@ -0,0 +1,273 @@
+---
+applyTo: '**/byok/**'
+description: Azure provider settings format and configuration guidelines
+---
+
+# Azure Provider Settings Configuration
+
+Guidelines for configuring Azure OpenAI models in the BYOK (Bring Your Own Key) system using the `chat.azureModels` VS Code setting.
+
+## Settings Format
+
+The `chat.azureModels` setting accepts a record of model configurations where each key is a unique model identifier and the value is a model configuration object.
+
+### Model Configuration Schema
+
+```typescript
+{
+  "chat.azureModels": {
+    "[modelId]": {
+      "name": string,                    // Display name for the model
+      "url": string,                     // Azure endpoint URL (base URL without API path)
+      "deploymentType": "completions" | "responses",  // API type (default: "completions")
+      "deploymentName": string,          // Azure deployment name (optional, defaults to modelId)
+      "apiVersion": string,              // Azure API version (optional, default: "2025-01-01-preview")
+      "temperature": number,             // Temperature setting (optional, 0.0-2.0)
+      "toolCalling": boolean,            // Whether model supports function/tool calling
+      "vision": boolean,                 // Whether model supports vision/image inputs
+      "maxInputTokens": number,          // Maximum input context window size
+      "maxOutputTokens": number,         // Maximum output tokens per response
+      "requiresAPIKey": boolean,         // Whether model requires API key (optional, default: true)
+      "thinking": boolean                // Whether model supports thinking/reasoning tokens (optional)
+    }
+  }
+}
+```
+
+## URL Resolution Behavior
+
+### Base URL Format
+The `url` field should contain the **base Azure endpoint URL** without the API path:
+
+```json
+// ✅ Correct - base URL only
+"url": "https://my-resource.openai.azure.com"
+"url": "https://my-resource.models.ai.azure.com"
+
+// ❌ Wrong - includes API path
+"url": "https://my-resource.openai.azure.com/openai/deployments/gpt-4/chat/completions"
+```
+
+### Deployment Types
+
+#### 1. Chat Completions API (`deploymentType: "completions"`)
+**Default deployment type.** Uses the standard Azure OpenAI Chat Completions endpoint.
+
+- **URL Pattern**: Deployment name is included in the URL path
+- **Azure OpenAI endpoints** (`openai.azure.com`):
+  - Resolved URL: `{url}/openai/deployments/{deploymentName}/chat/completions?api-version={apiVersion}`
+- **Azure AI endpoints** (`models.ai.azure.com`, `inference.ml.azure.com`):
+  - Resolved URL: `{url}/v1/chat/completions`
+
+**Example Configuration:**
+```json
+{
+  "chat.azureModels": {
+    "gpt-4-turbo": {
+      "name": "GPT-4 Turbo",
+      "url": "https://my-resource.openai.azure.com",
+      "deploymentType": "completions",
+      "deploymentName": "gpt-4-turbo-deployment",
+      "apiVersion": "2024-08-01-preview",
+      "toolCalling": true,
+      "vision": false,
+      "maxInputTokens": 128000,
+      "maxOutputTokens": 4096
+    }
+  }
+}
+```
+
+#### 2. Responses API (`deploymentType: "responses"`)
+Uses the Azure OpenAI Responses API for streaming responses with structured outputs.
+
+- **URL Pattern**: Deployment name is sent in the request body, NOT in URL
+- **Azure OpenAI endpoints** (`openai.azure.com`):
+  - Resolved URL: `{url}/openai/responses?api-version={apiVersion}`
+- **Azure AI endpoints** (`models.ai.azure.com`, `inference.ml.azure.com`):
+  - Resolved URL: `{url}/v1/responses?api-version={apiVersion}`
+- **Model Field**: The `deploymentName` is passed as the `model` field in the request body
+
+**Example Configuration:**
+```json
+{
+  "chat.azureModels": {
+    "o1-preview": {
+      "name": "OpenAI o1-preview",
+      "url": "https://my-resource.openai.azure.com",
+      "deploymentType": "responses",
+      "deploymentName": "o1-preview-deployment",
+      "apiVersion": "2025-01-01-preview",
+      "toolCalling": false,
+      "vision": true,
+      "thinking": true,
+      "maxInputTokens": 128000,
+      "maxOutputTokens": 32768
+    }
+  }
+}
+```
+
+### Deployment Name Behavior
+
+- **If `deploymentName` is provided**: Uses the specified deployment name
+- **If `deploymentName` is omitted**: Falls back to using `modelId` as the deployment name
+- **For Completions API**: Deployment name is included in the URL path
+- **For Responses API**: Deployment name is sent as the `model` field in the request body
+
+## Complete Configuration Examples
+
+### Example 1: Azure OpenAI - Chat Completions
+```json
+{
+  "chat.azureModels": {
+    "gpt-4": {
+      "name": "GPT-4",
+      "url": "https://my-eastus-resource.openai.azure.com",
+      "deploymentType": "completions",
+      "deploymentName": "gpt-4-prod",
+      "apiVersion": "2024-08-01-preview",
+      "temperature": 0.7,
+      "toolCalling": true,
+      "vision": false,
+      "maxInputTokens": 8192,
+      "maxOutputTokens": 4096,
+      "requiresAPIKey": true
+    }
+  }
+}
+```
+
+### Example 2: Azure AI - Responses API with Reasoning
+```json
+{
+  "chat.azureModels": {
+    "o1": {
+      "name": "OpenAI o1",
+      "url": "https://my-resource.models.ai.azure.com",
+      "deploymentType": "responses",
+      "deploymentName": "o1-deployment",
+      "apiVersion": "2025-01-01-preview",
+      "toolCalling": false,
+      "vision": true,
+      "thinking": true,
+      "maxInputTokens": 200000,
+      "maxOutputTokens": 100000,
+      "requiresAPIKey": true
+    }
+  }
+}
+```
+
+### Example 3: Multiple Models
+```json
+{
+  "chat.azureModels": {
+    "gpt-35-turbo": {
+      "name": "GPT-3.5 Turbo",
+      "url": "https://my-resource.openai.azure.com",
+      "deploymentType": "completions",
+      "deploymentName": "gpt-35-turbo-16k",
+      "toolCalling": true,
+      "vision": false,
+      "maxInputTokens": 16384,
+      "maxOutputTokens": 4096
+    },
+    "gpt-4-vision": {
+      "name": "GPT-4 Vision",
+      "url": "https://my-resource.openai.azure.com",
+      "deploymentType": "completions",
+      "deploymentName": "gpt-4-vision-preview",
+      "toolCalling": true,
+      "vision": true,
+      "maxInputTokens": 128000,
+      "maxOutputTokens": 4096
+    },
+    "o1-preview": {
+      "name": "OpenAI o1 Preview",
+      "url": "https://my-resource.openai.azure.com",
+      "deploymentType": "responses",
+      "deploymentName": "o1-preview",
+      "toolCalling": false,
+      "vision": true,
+      "thinking": true,
+      "maxInputTokens": 128000,
+      "maxOutputTokens": 32768
+    }
+  }
+}
+```
+
+## Implementation Details
+
+### URL Resolution Logic (`resolveAzureUrl`)
+
+The `resolveAzureUrl` function in `azureProvider.ts` handles URL construction:
+
+1. **Strips trailing slashes and `/v1` suffixes** from base URLs
+2. **Detects endpoint type** based on URL patterns:
+   - Azure AI: `models.ai.azure.com`, `inference.ml.azure.com`
+   - Azure OpenAI: `openai.azure.com`, `cognitiveservices.azure.com`
+3. **Applies deployment-specific URL patterns** based on `deploymentType`
+4. **Uses defaults** if optional fields are omitted:
+   - `deploymentType`: `"completions"`
+   - `deploymentName`: Falls back to `modelId`
+   - `apiVersion`: `"2025-01-01-preview"`
+
+### Model Info Resolution (`getModelInfo`)
+
+The `getModelInfo` method in `AzureBYOKModelProvider`:
+
+1. **Retrieves deployment configuration** from settings
+2. **For Responses API**: Overrides the model name to use `deploymentName` (sent in request body)
+3. **For Completions API**: Uses `deploymentName` in URL path
+4. **Sets `modelInfo.id`** to the deployment name for proper API routing
+5. **Configures `supported_endpoints`**:
+   - Responses API: `[ModelSupportedEndpoint.Responses]`
+   - Completions API: `[ModelSupportedEndpoint.ChatCompletions]`
+6. **Applies temperature** from configuration if specified
+
+### API Key Management
+
+- API keys are stored securely using VS Code's secrets storage
+- Keys are stored per-model using the pattern: `copilot-byok-Azure-{modelId}-api-key`
+- The `requiresAPIKey` field controls whether authentication is required (default: `true`)
+- Users configure API keys through the UI or via commands
+
+## Best Practices
+
+1. **Use explicit deployment names**: Always specify `deploymentName` to avoid confusion with model IDs
+2. **Match API versions**: Use the API version that matches your Azure deployment capabilities
+3. **Set accurate token limits**: Configure `maxInputTokens` and `maxOutputTokens` based on your deployment
+4. **Enable appropriate capabilities**: Set `toolCalling`, `vision`, and `thinking` flags based on model support
+5. **Test endpoint URLs**: Verify base URLs are correct and accessible before adding models
+6. **Group related models**: Use descriptive model IDs for easy identification in the UI
+7. **Document custom configurations**: Add comments in settings.json to explain non-standard configurations
+
+## Troubleshooting
+
+### Common Issues
+
+**Issue**: Model not appearing in selection
+- **Check**: Verify `requiresAPIKey` is set correctly and API key is configured
+- **Check**: Ensure all required fields are present in configuration
+
+**Issue**: 404 errors when using model
+- **Check**: Verify `deploymentName` matches your Azure deployment
+- **Check**: Ensure `url` points to the correct Azure resource
+- **Check**: Confirm `apiVersion` is supported by your deployment
+
+**Issue**: Unsupported endpoint errors
+- **Check**: Verify `deploymentType` matches your deployment's API type
+- **Check**: For reasoning models (o1), use `deploymentType: "responses"`
+
+**Issue**: Model name mismatches in API calls
+- **For Responses API**: Ensure `deploymentName` is set (it's sent as the model name)
+- **For Completions API**: Deployment name is in URL, not request body
+
+## Related Files
+
+- Implementation: `src/extension/byok/vscode-node/azureProvider.ts`
+- Base provider: `src/extension/byok/vscode-node/customOAIProvider.ts`
+- Configuration: `src/platform/configuration/common/configurationService.ts`
+- Storage: `src/extension/byok/vscode-node/byokStorageService.ts`

From 7b4ed96edb829579e23aef343b2ed4904e8ec5fe Mon Sep 17 00:00:00 2001
From: koureasstavros <koureasstavros@gmail.com>
Date: Mon, 3 Nov 2025 14:31:41 +0200
Subject: [PATCH 04/21] Removed debug logs from azure provider

---
 .../byok/vscode-node/azureProvider.ts         | 35 ++-----------------
 1 file changed, 2 insertions(+), 33 deletions(-)

diff --git a/src/extension/byok/vscode-node/azureProvider.ts b/src/extension/byok/vscode-node/azureProvider.ts
index 439b89961a..d2653617c8 100644
--- a/src/extension/byok/vscode-node/azureProvider.ts
+++ b/src/extension/byok/vscode-node/azureProvider.ts
@@ -115,7 +115,6 @@ export class AzureBYOKModelProvider extends CustomOAIBYOKModelProvider {
 	}
 
 	protected override resolveUrl(modelId: string, url: string): string {
-		this._logService?.info(`[Azure Debug] resolveUrl called for modelId: ${modelId}`);
 		try {
 			// Get model config to access deployment options
 			const modelConfig = this._configurationService?.getConfig(this.getConfigKey()) as Record<string, {
@@ -127,7 +126,6 @@ export class AzureBYOKModelProvider extends CustomOAIBYOKModelProvider {
 			}> | undefined;
 
 			const config = modelConfig?.[modelId];
-			this._logService?.info(`[Azure Debug] Model config for ${modelId}: ${JSON.stringify(config, null, 2)}`);
 
 			const options: AzureUrlOptions | undefined = config ? {
 				deploymentType: config.deploymentType,
@@ -136,25 +134,19 @@ export class AzureBYOKModelProvider extends CustomOAIBYOKModelProvider {
 			} : undefined;
 
 			const resolvedUrl = resolveAzureUrl(modelId, url, options, this._logService);
-			this._logService?.info(`[Azure Debug] Final resolved URL: ${resolvedUrl}`);
 			return resolvedUrl;
 		} catch (error) {
-			this._logService?.error(`AzureBYOKModelProvider: Error resolving URL for model ${modelId}:`, error);
 			// Fallback to basic URL resolution
 			return resolveAzureUrl(modelId, url, undefined, this._logService);
 		}
 	}
 
 	protected override async getModelInfo(modelId: string, apiKey: string | undefined, modelCapabilities?: BYOKModelCapabilities): Promise<IChatModelInformation> {
-		this._logService?.info(`[Azure Debug] getModelInfo called for modelId: ${modelId}`);
-		this._logService?.info(`[Azure Debug] Initial modelCapabilities: ${JSON.stringify(modelCapabilities, null, 2)}`);
 		try {
 			// Get model config to check deployment type and deployment name
 			const configKey = this.getConfigKey();
-			this._logService?.info(`[Azure Debug] Config key: ${configKey}`);
 
 			const modelConfig = this._configurationService?.getConfig(configKey);
-			this._logService?.info(`[Azure Debug] Raw modelConfig type: ${typeof modelConfig}, value: ${JSON.stringify(modelConfig, null, 2)}`);
 
 			// Safely access the model-specific config
 			let config: { deploymentType?: 'completions' | 'responses'; deploymentName?: string; temperature?: number } | undefined;
@@ -162,29 +154,11 @@ export class AzureBYOKModelProvider extends CustomOAIBYOKModelProvider {
 				config = (modelConfig as Record<string, any>)[modelId];
 			}
 
-			this._logService?.info(`[Azure Debug] Model config for ${modelId}: ${JSON.stringify(config, null, 2)}`);
-
 			const deploymentType = config?.deploymentType || 'completions';
 			// If no deployment name is provided, use modelId as fallback
 			const deploymentName = config?.deploymentName || modelId;
 
-			this._logService?.info(`[Azure Debug] Deployment info - type: ${deploymentType}, name: ${deploymentName}`);
-
-			// For Azure Responses API, the 'model' field in the request body must be the deployment name
-			// We override the modelCapabilities.name to be the deployment name so it gets used correctly
-			let updatedCapabilities = modelCapabilities;
-			if (deploymentType === 'responses' && modelCapabilities) {
-				updatedCapabilities = {
-					...modelCapabilities,
-					name: deploymentName
-				};
-				this._logService?.info(`[Azure Debug] RESPONSES API: Overriding model name from '${modelCapabilities.name}' to '${deploymentName}'`);
-				this._logService?.info(`[Azure Debug] Updated capabilities: ${JSON.stringify(updatedCapabilities, null, 2)}`);
-			} else {
-				this._logService?.info(`[Azure Debug] COMPLETIONS API: Using capabilities as-is (deployment name in URL)`);
-			}
-
-			const modelInfo = await super.getModelInfo(modelId, apiKey, updatedCapabilities);
+			const modelInfo = await super.getModelInfo(modelId, apiKey, modelCapabilities);
 
 			// Always set modelInfo.id to deployment name for Azure deployments
 			modelInfo.id = deploymentName;
@@ -192,21 +166,16 @@ export class AzureBYOKModelProvider extends CustomOAIBYOKModelProvider {
 			// Set temperature from config if specified
 			if (config?.temperature !== undefined) {
 				modelInfo.temperature = config.temperature;
-				this._logService?.info(`[Azure Debug] Set temperature from config: ${config.temperature}`);
+				this._logService?.info(`[Warning] Set temperature from config: ${config.temperature}`);
 			}
 
 			// Set supported endpoints based on deployment type
 			if (deploymentType === 'responses') {
 				modelInfo.supported_endpoints = [ModelSupportedEndpoint.Responses];
-				this._logService?.info(`[Azure Debug] Set supported_endpoints to: [Responses]`);
 			} else {
 				// For completions API, only support chat completions
 				modelInfo.supported_endpoints = [ModelSupportedEndpoint.ChatCompletions];
-				this._logService?.info(`[Azure Debug] Set supported_endpoints to: [ChatCompletions]`);
 			}
-			this._logService?.info(`[Azure Debug] Set modelInfo.id to deployment name: '${deploymentName}'`);
-			this._logService?.info(`[Azure Debug] Final modelInfo.id: ${modelInfo.id}`);
-			this._logService?.info(`[Azure Debug] Final modelInfo.supported_endpoints: ${JSON.stringify(modelInfo.supported_endpoints)}`);
 
 			return modelInfo;
 		} catch (error) {

From a1428db36fc10a21bb9923baa0ef6666a82c2061 Mon Sep 17 00:00:00 2001
From: koureasstavros <koureasstavros@gmail.com>
Date: Mon, 3 Nov 2025 14:40:36 +0200
Subject: [PATCH 05/21] Fixed Copilot reviewe suggestions

---
 src/extension/byok/node/openAIEndpoint.ts       | 17 +++++++++++------
 src/extension/byok/vscode-node/azureProvider.ts | 11 +----------
 .../codeBlocks/node/codeBlockProcessor.ts       |  2 +-
 3 files changed, 13 insertions(+), 17 deletions(-)

diff --git a/src/extension/byok/node/openAIEndpoint.ts b/src/extension/byok/node/openAIEndpoint.ts
index a8e0182c73..53d18e3140 100644
--- a/src/extension/byok/node/openAIEndpoint.ts
+++ b/src/extension/byok/node/openAIEndpoint.ts
@@ -291,17 +291,22 @@ export class OpenAIEndpoint extends ChatEndpoint {
 		}
 
 		if (body) {
+			// Apply temperature from config if specified (before thinking model checks)
+			if (this._temperature !== undefined) {
+				body.temperature = this._temperature;
+			}
+
 			if (this.modelMetadata.capabilities.supports.thinking) {
+				// Thinking models don't support temperature parameter
 				delete body.temperature;
+				// Thinking models use max_completion_tokens instead of max_tokens
 				body['max_completion_tokens'] = body.max_tokens;
 				delete body.max_tokens;
+			} else {
+				// For non-thinking models, removing max_tokens defaults to the maximum which is what we want for BYOK
+				delete body.max_tokens;
 			}
-			// Apply temperature from config if specified (and not a thinking model)
-			if (this._temperature !== undefined && !this.modelMetadata.capabilities.supports.thinking) {
-				body.temperature = this._temperature;
-			}
-			// Removing max tokens defaults to the maximum which is what we want for BYOK
-			delete body.max_tokens;
+
 			if (!this.useResponsesApi && body.stream) {
 				body['stream_options'] = { 'include_usage': true };
 			}
diff --git a/src/extension/byok/vscode-node/azureProvider.ts b/src/extension/byok/vscode-node/azureProvider.ts
index d2653617c8..7d74bf819a 100644
--- a/src/extension/byok/vscode-node/azureProvider.ts
+++ b/src/extension/byok/vscode-node/azureProvider.ts
@@ -19,12 +19,9 @@ export interface AzureUrlOptions {
 }
 
 export function resolveAzureUrl(modelId: string, url: string, options?: AzureUrlOptions, logService?: ILogService): string {
-	logService?.info(`[Azure Debug] resolveAzureUrl called - modelId: ${modelId}, url: ${url}`);
-	logService?.info(`[Azure Debug] options: ${JSON.stringify(options, null, 2)}`);
 
 	// The fully resolved url was already passed in
 	if (hasExplicitApiPath(url)) {
-		logService?.info(`[Azure Debug] URL has explicit API path, returning as-is: ${url}`);
 		return url;
 	}
 
@@ -43,14 +40,10 @@ export function resolveAzureUrl(modelId: string, url: string, options?: AzureUrl
 	const deploymentType = options?.deploymentType || 'completions';
 	const apiVersion = options?.apiVersion || '2025-01-01-preview';
 
-	logService?.info(`[Azure Debug] Resolved values - deploymentName: ${deploymentName}, deploymentType: ${deploymentType}, apiVersion: ${apiVersion}`);
-
 	// Determine if this is an Azure OpenAI endpoint (requires deployment name in path)
 	const isAzureOpenAIEndpointv1 = url.includes('models.ai.azure.com') || url.includes('inference.ml.azure.com');
 	const isAzureOpenAIEndpoint = url.includes('openai.azure.com') || url.includes('cognitiveservices.azure.com');
 
-	logService?.info(`[Azure Debug] Endpoint detection - v1: ${isAzureOpenAIEndpointv1}, standard: ${isAzureOpenAIEndpoint}`);
-
 	if (deploymentType === 'responses') {
 		// Handle Responses API
 		// Deployment name is passed in the request body as 'model' parameter, not in URL
@@ -62,8 +55,7 @@ export function resolveAzureUrl(modelId: string, url: string, options?: AzureUrl
 		} else {
 			throw new Error(`Unrecognized Azure deployment URL for Responses API: ${url}`);
 		}
-		logService?.info(`[Azure Debug] Responses API URL resolved to: ${resolvedUrl}`);
-		logService?.info(`[Azure Debug] NOTE: Deployment name '${deploymentName}' will be sent in request body as 'model' field`);
+
 		return resolvedUrl;
 	} else if (deploymentType === 'completions') {
 		// Handle Chat Completions API (default)
@@ -77,7 +69,6 @@ export function resolveAzureUrl(modelId: string, url: string, options?: AzureUrl
 		} else {
 			throw new Error(`Unrecognized Azure deployment URL: ${url}`);
 		}
-		logService?.info(`[Azure Debug] Completions API URL resolved to: ${resolvedUrl}`);
 		return resolvedUrl;
 	} else {
 		throw new Error(`Invalid deployment type specified for model ${modelId}: ${deploymentType}`);
diff --git a/src/extension/codeBlocks/node/codeBlockProcessor.ts b/src/extension/codeBlocks/node/codeBlockProcessor.ts
index c5671f5fd8..63c25c0246 100644
--- a/src/extension/codeBlocks/node/codeBlockProcessor.ts
+++ b/src/extension/codeBlocks/node/codeBlockProcessor.ts
@@ -106,7 +106,7 @@ export class CodeBlockTrackingChatResponseStream implements ChatResponseStream {
 		const method = this._wrapped[methodName];
 		if (typeof method !== 'function') {
 			// Return a no-op function for missing methods
-			return ((() => { }) as any);
+			return (() => { }) as any;
 		}
 		return ((...args: any[]) => {
 			this._codeBlockProcessor.flush();

From edaa621ee37e33349edd7ee9a599f6d4b58d692a Mon Sep 17 00:00:00 2001
From: koureasstavros <koureasstavros@gmail.com>
Date: Mon, 3 Nov 2025 15:19:52 +0200
Subject: [PATCH 06/21] Fixed Copilot reviewe suggestions

---
 src/extension/byok/vscode-node/azureProvider.ts     | 13 ++++++-------
 .../configuration/common/configurationService.ts    |  2 +-
 2 files changed, 7 insertions(+), 8 deletions(-)

diff --git a/src/extension/byok/vscode-node/azureProvider.ts b/src/extension/byok/vscode-node/azureProvider.ts
index 7d74bf819a..784fbe0704 100644
--- a/src/extension/byok/vscode-node/azureProvider.ts
+++ b/src/extension/byok/vscode-node/azureProvider.ts
@@ -13,9 +13,9 @@ import { IBYOKStorageService } from './byokStorageService';
 import { CustomOAIBYOKModelProvider, hasExplicitApiPath } from './customOAIProvider';
 
 export interface AzureUrlOptions {
-	deploymentType?: 'completions' | 'responses';
-	deploymentName?: string;
-	apiVersion?: string;
+	deploymentType: 'completions' | 'responses';
+	deploymentName: string;
+	apiVersion: string;
 }
 
 export function resolveAzureUrl(modelId: string, url: string, options?: AzureUrlOptions, logService?: ILogService): string {
@@ -119,9 +119,9 @@ export class AzureBYOKModelProvider extends CustomOAIBYOKModelProvider {
 			const config = modelConfig?.[modelId];
 
 			const options: AzureUrlOptions | undefined = config ? {
-				deploymentType: config.deploymentType,
-				deploymentName: config.deploymentName,
-				apiVersion: config.apiVersion
+				deploymentType: config.deploymentType || 'completions',
+				deploymentName: config.deploymentName || modelId,
+				apiVersion: config.apiVersion || '2025-01-01-preview'
 			} : undefined;
 
 			const resolvedUrl = resolveAzureUrl(modelId, url, options, this._logService);
@@ -157,7 +157,6 @@ export class AzureBYOKModelProvider extends CustomOAIBYOKModelProvider {
 			// Set temperature from config if specified
 			if (config?.temperature !== undefined) {
 				modelInfo.temperature = config.temperature;
-				this._logService?.info(`[Warning] Set temperature from config: ${config.temperature}`);
 			}
 
 			// Set supported endpoints based on deployment type
diff --git a/src/platform/configuration/common/configurationService.ts b/src/platform/configuration/common/configurationService.ts
index 2ca0159139..99bff89936 100644
--- a/src/platform/configuration/common/configurationService.ts
+++ b/src/platform/configuration/common/configurationService.ts
@@ -812,7 +812,7 @@ export namespace ConfigKey {
 	/** BYOK  */
 	export const OllamaEndpoint = defineSetting<string>('chat.byok.ollamaEndpoint', 'http://localhost:11434');
 	export const AzureModels = defineSetting<Record<string, { name: string; url: string; deploymentType: 'completions' | 'responses'; deploymentName: string; apiVersion: string; toolCalling: boolean; vision: boolean; maxInputTokens: number; maxOutputTokens: number; requiresAPIKey?: boolean; thinking?: boolean }>>('chat.azureModels', {});
-	export const CustomOAIModels = defineSetting<Record<string, { name: string; url: string; toolCalling: boolean; vision: boolean; maxInputTokens: number; maxOutputTokens: number; requiresAPIKey?: boolean; thinking?: boolean; requestHeaders?: Record<string, string> }>>('chat.customOAIModels', {});
+	export const CustomOAIModels = defineSetting<Record<string, { name: string; url: string; toolCalling: boolean; vision: boolean; maxInputTokens: number; maxOutputTokens: number; requiresAPIKey?: boolean; thinking?: boolean; temperature?: number; requestHeaders?: Record<string, string> }>>('chat.customOAIModels', {});
 	export const AutoFixDiagnostics = defineExpSetting<boolean>('chat.agent.autoFix', true);
 	export const NotebookFollowCellExecution = defineSetting<boolean>('chat.notebook.followCellExecution.enabled', false);
 	export const UseAlternativeNESNotebookFormat = defineExpSetting<boolean>('chat.notebook.enhancedNextEditSuggestions.enabled', false);

From a0f891f2670d6f0dc3ae39b2f046143cc617b0e8 Mon Sep 17 00:00:00 2001
From: koureasstavros <koureasstavros@gmail.com>
Date: Mon, 3 Nov 2025 16:06:07 +0200
Subject: [PATCH 07/21] Fixed Copilot reviews suggestions

---
 package.json                                  |  4 +++-
 .../byok/vscode-node/azureProvider.ts         | 20 ++++++++++---------
 2 files changed, 14 insertions(+), 10 deletions(-)

diff --git a/package.json b/package.json
index d2be7301cf..22e17f467d 100644
--- a/package.json
+++ b/package.json
@@ -3070,7 +3070,9 @@
 								},
 								"temperature": {
 									"type": "number",
-									"description": "Temperature parameter for model responses (typically 0.0 to 2.0)"
+									"description": "Temperature parameter for model responses (typically 0.0 to 2.0)",
+									"minimum": 0,
+									"maximum": 2
 								},
 								"requestHeaders": {
 									"type": "object",
diff --git a/src/extension/byok/vscode-node/azureProvider.ts b/src/extension/byok/vscode-node/azureProvider.ts
index 784fbe0704..f338cb5ed7 100644
--- a/src/extension/byok/vscode-node/azureProvider.ts
+++ b/src/extension/byok/vscode-node/azureProvider.ts
@@ -18,6 +18,14 @@ export interface AzureUrlOptions {
 	apiVersion: string;
 }
 
+interface AzureModelConfig {
+	deploymentType?: 'completions' | 'responses';
+	deploymentName?: string;
+	apiVersion?: string;
+	temperature?: number;
+	url: string;
+}
+
 export function resolveAzureUrl(modelId: string, url: string, options?: AzureUrlOptions, logService?: ILogService): string {
 
 	// The fully resolved url was already passed in
@@ -108,13 +116,7 @@ export class AzureBYOKModelProvider extends CustomOAIBYOKModelProvider {
 	protected override resolveUrl(modelId: string, url: string): string {
 		try {
 			// Get model config to access deployment options
-			const modelConfig = this._configurationService?.getConfig(this.getConfigKey()) as Record<string, {
-				deploymentType?: 'completions' | 'responses';
-				deploymentName?: string;
-				apiVersion?: string;
-				temperature?: number;
-				url: string;
-			}> | undefined;
+			const modelConfig = this._configurationService?.getConfig(this.getConfigKey()) as Record<string, AzureModelConfig> | undefined;
 
 			const config = modelConfig?.[modelId];
 
@@ -140,9 +142,9 @@ export class AzureBYOKModelProvider extends CustomOAIBYOKModelProvider {
 			const modelConfig = this._configurationService?.getConfig(configKey);
 
 			// Safely access the model-specific config
-			let config: { deploymentType?: 'completions' | 'responses'; deploymentName?: string; temperature?: number } | undefined;
+			let config: AzureModelConfig | undefined;
 			if (modelConfig && typeof modelConfig === 'object' && modelId in modelConfig) {
-				config = (modelConfig as Record<string, any>)[modelId];
+				config = (modelConfig as Record<string, AzureModelConfig>)[modelId];
 			}
 
 			const deploymentType = config?.deploymentType || 'completions';

From a4227d36fe876c672125e115399d0053ff0db4e8 Mon Sep 17 00:00:00 2001
From: koureasstavros <koureasstavros@gmail.com>
Date: Tue, 4 Nov 2025 10:54:33 +0200
Subject: [PATCH 08/21] Fixed Copilot reviews suggestions

---
 src/extension/byok/vscode-node/azureProvider.ts     | 2 +-
 src/extension/codeBlocks/node/codeBlockProcessor.ts | 4 +++-
 2 files changed, 4 insertions(+), 2 deletions(-)

diff --git a/src/extension/byok/vscode-node/azureProvider.ts b/src/extension/byok/vscode-node/azureProvider.ts
index f338cb5ed7..97b0d59d87 100644
--- a/src/extension/byok/vscode-node/azureProvider.ts
+++ b/src/extension/byok/vscode-node/azureProvider.ts
@@ -129,7 +129,7 @@ export class AzureBYOKModelProvider extends CustomOAIBYOKModelProvider {
 			const resolvedUrl = resolveAzureUrl(modelId, url, options, this._logService);
 			return resolvedUrl;
 		} catch (error) {
-			// Fallback to basic URL resolution
+			this._logService?.error(`AzureBYOKModelProvider: Error resolving URL for model ${modelId}, falling back to basic resolution:`, error);
 			return resolveAzureUrl(modelId, url, undefined, this._logService);
 		}
 	}
diff --git a/src/extension/codeBlocks/node/codeBlockProcessor.ts b/src/extension/codeBlocks/node/codeBlockProcessor.ts
index 63c25c0246..7e52e585c5 100644
--- a/src/extension/codeBlocks/node/codeBlockProcessor.ts
+++ b/src/extension/codeBlocks/node/codeBlockProcessor.ts
@@ -9,6 +9,7 @@ import { createFilepathRegexp, mdCodeBlockLangToLanguageId } from '../../../util
 import { CharCode } from '../../../util/vs/base/common/charCode';
 import { isFalsyOrWhitespace, splitLinesIncludeSeparators } from '../../../util/vs/base/common/strings';
 
+import { ILogService } from '../../../platform/log/common/logService';
 import { IPromptPathRepresentationService } from '../../../platform/prompts/common/promptPathRepresentationService';
 import { ChatResponseCodeblockUriPart, ChatResponseMarkdownPart, ChatResponseMarkdownWithVulnerabilitiesPart, MarkdownString } from '../../../vscodeTypes';
 import { CodeBlock } from '../../prompt/common/conversation';
@@ -38,6 +39,7 @@ export class CodeBlockTrackingChatResponseStream implements ChatResponseStream {
 		private readonly _wrapped: ChatResponseStream,
 		codeblocksRepresentEdits: boolean | undefined,
 		@IPromptPathRepresentationService _promptPathRepresentationService: IPromptPathRepresentationService,
+		@ILogService private readonly _logService: ILogService,
 	) {
 		let uriReportedForIndex = -1;
 		this._codeBlockProcessor = new CodeBlockProcessor(
@@ -105,7 +107,7 @@ export class CodeBlockTrackingChatResponseStream implements ChatResponseStream {
 	private forward<K extends keyof ChatResponseStream>(methodName: K): ChatResponseStream[K] {
 		const method = this._wrapped[methodName];
 		if (typeof method !== 'function') {
-			// Return a no-op function for missing methods
+			this._logService.warn(`[CodeBlockTrackingChatResponseStream] Method '${String(methodName)}' does not exist on the wrapped ChatResponseStream.`);
 			return (() => { }) as any;
 		}
 		return ((...args: any[]) => {

From 73eda6534e856085bf7f7f3d998d2f0730afc3c6 Mon Sep 17 00:00:00 2001
From: koureasstavros <koureasstavros@gmail.com>
Date: Tue, 4 Nov 2025 11:11:52 +0200
Subject: [PATCH 09/21] Fixed Copilot reviews suggestions

---
 src/extension/byok/vscode-node/azureProvider.ts     | 5 -----
 src/extension/codeBlocks/node/codeBlockProcessor.ts | 2 +-
 2 files changed, 1 insertion(+), 6 deletions(-)

diff --git a/src/extension/byok/vscode-node/azureProvider.ts b/src/extension/byok/vscode-node/azureProvider.ts
index 97b0d59d87..5f78731445 100644
--- a/src/extension/byok/vscode-node/azureProvider.ts
+++ b/src/extension/byok/vscode-node/azureProvider.ts
@@ -102,11 +102,6 @@ export class AzureBYOKModelProvider extends CustomOAIBYOKModelProvider {
 		);
 		// Override the instance properties
 		this.providerName = AzureBYOKModelProvider.providerName;
-
-		// Bind methods to ensure proper context
-		this.resolveUrl = this.resolveUrl.bind(this);
-		this.getModelInfo = this.getModelInfo.bind(this);
-		this.getConfigKey = this.getConfigKey.bind(this);
 	}
 
 	protected override getConfigKey() {
diff --git a/src/extension/codeBlocks/node/codeBlockProcessor.ts b/src/extension/codeBlocks/node/codeBlockProcessor.ts
index 7e52e585c5..b6419f948a 100644
--- a/src/extension/codeBlocks/node/codeBlockProcessor.ts
+++ b/src/extension/codeBlocks/node/codeBlockProcessor.ts
@@ -108,7 +108,7 @@ export class CodeBlockTrackingChatResponseStream implements ChatResponseStream {
 		const method = this._wrapped[methodName];
 		if (typeof method !== 'function') {
 			this._logService.warn(`[CodeBlockTrackingChatResponseStream] Method '${String(methodName)}' does not exist on the wrapped ChatResponseStream.`);
-			return (() => { }) as any;
+			return (() => undefined) as any;
 		}
 		return ((...args: any[]) => {
 			this._codeBlockProcessor.flush();

From fa45fb2462e20131615dc9b3767678714b59f1a3 Mon Sep 17 00:00:00 2001
From: koureasstavros <koureasstavros@gmail.com>
Date: Tue, 4 Nov 2025 11:32:35 +0200
Subject: [PATCH 10/21] Fixed Copilot reviews suggestions

---
 .../byok/vscode-node/azureProvider.ts         |  6 ++---
 .../byok/vscode-node/customOAIProvider.ts     | 23 ++++++++++++++++---
 2 files changed, 23 insertions(+), 6 deletions(-)

diff --git a/src/extension/byok/vscode-node/azureProvider.ts b/src/extension/byok/vscode-node/azureProvider.ts
index 5f78731445..671eb8c428 100644
--- a/src/extension/byok/vscode-node/azureProvider.ts
+++ b/src/extension/byok/vscode-node/azureProvider.ts
@@ -26,7 +26,7 @@ interface AzureModelConfig {
 	url: string;
 }
 
-export function resolveAzureUrl(modelId: string, url: string, options?: AzureUrlOptions, logService?: ILogService): string {
+export function resolveAzureUrl(modelId: string, url: string, options?: AzureUrlOptions): string {
 
 	// The fully resolved url was already passed in
 	if (hasExplicitApiPath(url)) {
@@ -121,11 +121,11 @@ export class AzureBYOKModelProvider extends CustomOAIBYOKModelProvider {
 				apiVersion: config.apiVersion || '2025-01-01-preview'
 			} : undefined;
 
-			const resolvedUrl = resolveAzureUrl(modelId, url, options, this._logService);
+			const resolvedUrl = resolveAzureUrl(modelId, url, options);
 			return resolvedUrl;
 		} catch (error) {
 			this._logService?.error(`AzureBYOKModelProvider: Error resolving URL for model ${modelId}, falling back to basic resolution:`, error);
-			return resolveAzureUrl(modelId, url, undefined, this._logService);
+			return resolveAzureUrl(modelId, url, undefined);
 		}
 	}
 
diff --git a/src/extension/byok/vscode-node/customOAIProvider.ts b/src/extension/byok/vscode-node/customOAIProvider.ts
index bdc2a36a37..feb5284d92 100644
--- a/src/extension/byok/vscode-node/customOAIProvider.ts
+++ b/src/extension/byok/vscode-node/customOAIProvider.ts
@@ -44,10 +44,27 @@ export function hasExplicitApiPath(url: string): boolean {
 	return url.includes('/responses') || url.includes('/chat/completions');
 }
 
+interface UserModelConfig {
+	requestHeaders?: Record<string, string>;
+	name: string;
+	url: string;
+	deploymentType?: 'completions' | 'responses';
+	deploymentName?: string;
+	apiVersion?: string;
+	maxInputTokens: number;
+	maxOutputTokens: number;
+	requiresAPIKey: boolean;
+	toolCalling: boolean;
+	editTools?: EndpointEditToolName[];
+	vision: boolean;
+	thinking?: boolean;
+	temperature?: number;
+}
+
 interface CustomOAIModelInfo extends LanguageModelChatInformation {
+	requestHeaders?: Record<string, string>;
 	url: string;
 	thinking: boolean;
-	requestHeaders?: Record<string, string>;
 	temperature?: number;
 }
 
@@ -87,8 +104,8 @@ export class CustomOAIBYOKModelProvider implements BYOKModelProvider<CustomOAIMo
 		return modelInfo;
 	}
 
-	private getUserModelConfig(): Record<string, { name: string; url: string; toolCalling: boolean; vision: boolean; maxInputTokens: number; maxOutputTokens: number; requiresAPIKey: boolean; thinking?: boolean; editTools?: EndpointEditToolName[]; requestHeaders?: Record<string, string>; deploymentType?: 'completions' | 'responses'; deploymentName?: string; apiVersion?: string; temperature?: number }> {
-		const modelConfig = this._configurationService.getConfig(this.getConfigKey()) as Record<string, { name: string; url: string; toolCalling: boolean; vision: boolean; maxInputTokens: number; maxOutputTokens: number; requiresAPIKey: boolean; thinking?: boolean; editTools?: EndpointEditToolName[]; requestHeaders?: Record<string, string>; deploymentType?: 'completions' | 'responses'; deploymentName?: string; apiVersion?: string; temperature?: number }>;
+	private getUserModelConfig(): Record<string, UserModelConfig> {
+		const modelConfig = this._configurationService.getConfig(this.getConfigKey()) as Record<string, UserModelConfig>;
 		return modelConfig;
 	}
 

From cc11c74d7e82bfcc9228eacf1ade397ce052ac85 Mon Sep 17 00:00:00 2001
From: koureasstavros <koureasstavros@gmail.com>
Date: Tue, 4 Nov 2025 11:59:51 +0200
Subject: [PATCH 11/21] Fixed Copilot reviews suggestions

---
 .../byok/vscode-node/azureProvider.ts         |  2 +-
 .../byok/vscode-node/customOAIProvider.ts     | 30 +++++++++----------
 2 files changed, 16 insertions(+), 16 deletions(-)

diff --git a/src/extension/byok/vscode-node/azureProvider.ts b/src/extension/byok/vscode-node/azureProvider.ts
index 671eb8c428..71a6fdfed2 100644
--- a/src/extension/byok/vscode-node/azureProvider.ts
+++ b/src/extension/byok/vscode-node/azureProvider.ts
@@ -148,7 +148,7 @@ export class AzureBYOKModelProvider extends CustomOAIBYOKModelProvider {
 
 			const modelInfo = await super.getModelInfo(modelId, apiKey, modelCapabilities);
 
-			// Always set modelInfo.id to deployment name for Azure deployments
+			// Set modelInfo.id to deployment name (or modelId if no deployment name configured)
 			modelInfo.id = deploymentName;
 
 			// Set temperature from config if specified
diff --git a/src/extension/byok/vscode-node/customOAIProvider.ts b/src/extension/byok/vscode-node/customOAIProvider.ts
index feb5284d92..7ac98615b7 100644
--- a/src/extension/byok/vscode-node/customOAIProvider.ts
+++ b/src/extension/byok/vscode-node/customOAIProvider.ts
@@ -45,7 +45,6 @@ export function hasExplicitApiPath(url: string): boolean {
 }
 
 interface UserModelConfig {
-	requestHeaders?: Record<string, string>;
 	name: string;
 	url: string;
 	deploymentType?: 'completions' | 'responses';
@@ -59,13 +58,14 @@ interface UserModelConfig {
 	vision: boolean;
 	thinking?: boolean;
 	temperature?: number;
+	requestHeaders?: Record<string, string>;
 }
 
 interface CustomOAIModelInfo extends LanguageModelChatInformation {
-	requestHeaders?: Record<string, string>;
 	url: string;
 	thinking: boolean;
 	temperature?: number;
+	requestHeaders?: Record<string, string>;
 }
 
 export class CustomOAIBYOKModelProvider implements BYOKModelProvider<CustomOAIModelInfo> {
@@ -125,14 +125,14 @@ export class CustomOAIBYOKModelProvider implements BYOKModelProvider<CustomOAIMo
 			models[modelId] = {
 				name: modelInfo.name,
 				url: resolvedUrl,
-				toolCalling: modelInfo.toolCalling,
-				vision: modelInfo.vision,
 				maxInputTokens: modelInfo.maxInputTokens,
 				maxOutputTokens: modelInfo.maxOutputTokens,
-				thinking: modelInfo.thinking,
+				toolCalling: modelInfo.toolCalling,
 				editTools: modelInfo.editTools,
+				vision: modelInfo.vision,
+				thinking: modelInfo.thinking,
+				temperature: modelInfo.temperature,
 				requestHeaders: modelInfo.requestHeaders ? { ...modelInfo.requestHeaders } : undefined,
-				temperature: modelInfo.temperature
 			};
 		}
 		return models;
@@ -164,8 +164,8 @@ export class CustomOAIBYOKModelProvider implements BYOKModelProvider<CustomOAIMo
 	private createModelInfo(id: string, capabilities: BYOKKnownModels[string]): CustomOAIModelInfo {
 		const baseInfo: CustomOAIModelInfo = {
 			id,
-			url: capabilities.url || '',
 			name: capabilities.name,
+			url: capabilities.url || '',
 			detail: this.providerName,
 			version: '1.0.0',
 			maxOutputTokens: capabilities.maxOutputTokens,
@@ -178,8 +178,8 @@ export class CustomOAIBYOKModelProvider implements BYOKModelProvider<CustomOAIMo
 				editTools: capabilities.editTools
 			},
 			thinking: capabilities.thinking || false,
-			requestHeaders: capabilities.requestHeaders,
 			temperature: capabilities.temperature,
+			requestHeaders: capabilities.requestHeaders,
 		};
 		return baseInfo;
 	}
@@ -210,16 +210,16 @@ export class CustomOAIBYOKModelProvider implements BYOKModelProvider<CustomOAIMo
 			}
 		}
 		const modelInfo = await this.getModelInfo(model.id, apiKey, {
+			name: model.name,
+			url: model.url,
 			maxInputTokens: model.maxInputTokens,
 			maxOutputTokens: model.maxOutputTokens,
 			toolCalling: !!model.capabilities?.toolCalling || false,
+			editTools: model.capabilities.editTools?.filter(isEndpointEditToolName),
 			vision: !!model.capabilities?.imageInput || false,
-			name: model.name,
-			url: model.url,
 			thinking: model.thinking,
-			editTools: model.capabilities.editTools?.filter(isEndpointEditToolName),
-			requestHeaders: model.requestHeaders,
 			temperature: model.temperature,
+			requestHeaders: model.requestHeaders,
 		});
 		const openAIChatEndpoint = this._instantiationService.createInstance(OpenAIEndpoint, modelInfo, apiKey ?? '', model.url);
 		return this._lmWrapper.provideLanguageModelResponse(openAIChatEndpoint, messages, options, options.requestInitiator, progress, token);
@@ -237,15 +237,15 @@ export class CustomOAIBYOKModelProvider implements BYOKModelProvider<CustomOAIMo
 		}
 
 		const modelInfo = await this.getModelInfo(model.id, apiKey, {
+			name: model.name,
+			url: model.url,
 			maxInputTokens: model.maxInputTokens,
 			maxOutputTokens: model.maxOutputTokens,
 			toolCalling: !!model.capabilities?.toolCalling || false,
 			vision: !!model.capabilities?.imageInput || false,
-			name: model.name,
-			url: model.url,
 			thinking: model.thinking,
-			requestHeaders: model.requestHeaders,
 			temperature: model.temperature,
+			requestHeaders: model.requestHeaders,
 		});
 		const openAIChatEndpoint = this._instantiationService.createInstance(OpenAIEndpoint, modelInfo, apiKey ?? '', model.url);
 		return this._lmWrapper.provideTokenCount(openAIChatEndpoint, text);

From e7fa971c3204d5f4c92ae428923d280bd7042a2f Mon Sep 17 00:00:00 2001
From: koureasstavros <koureasstavros@gmail.com>
Date: Tue, 4 Nov 2025 12:04:20 +0200
Subject: [PATCH 12/21] Fixed Copilot reviews suggestions

---
 src/extension/byok/node/openAIEndpoint.ts | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/src/extension/byok/node/openAIEndpoint.ts b/src/extension/byok/node/openAIEndpoint.ts
index 53d18e3140..da70c75734 100644
--- a/src/extension/byok/node/openAIEndpoint.ts
+++ b/src/extension/byok/node/openAIEndpoint.ts
@@ -301,12 +301,12 @@ export class OpenAIEndpoint extends ChatEndpoint {
 				delete body.temperature;
 				// Thinking models use max_completion_tokens instead of max_tokens
 				body['max_completion_tokens'] = body.max_tokens;
-				delete body.max_tokens;
-			} else {
-				// For non-thinking models, removing max_tokens defaults to the maximum which is what we want for BYOK
-				delete body.max_tokens;
 			}
 
+			// For thinking models, replace max_tokens with max_completion_tokens
+			// For non-thinking models, removing max_tokens defaults to the maximum which is what we want for BYOK
+			delete body.max_tokens;
+
 			if (!this.useResponsesApi && body.stream) {
 				body['stream_options'] = { 'include_usage': true };
 			}

From 13b4b3f29968a7e13ce8e8d8e53764f557c86c4e Mon Sep 17 00:00:00 2001
From: koureasstavros <koureasstavros@gmail.com>
Date: Tue, 4 Nov 2025 12:07:48 +0200
Subject: [PATCH 13/21] Fixed Copilot reviews suggestions

---
 src/extension/codeBlocks/node/codeBlockProcessor.ts | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/src/extension/codeBlocks/node/codeBlockProcessor.ts b/src/extension/codeBlocks/node/codeBlockProcessor.ts
index b6419f948a..724d5fba10 100644
--- a/src/extension/codeBlocks/node/codeBlockProcessor.ts
+++ b/src/extension/codeBlocks/node/codeBlockProcessor.ts
@@ -108,12 +108,14 @@ export class CodeBlockTrackingChatResponseStream implements ChatResponseStream {
 		const method = this._wrapped[methodName];
 		if (typeof method !== 'function') {
 			this._logService.warn(`[CodeBlockTrackingChatResponseStream] Method '${String(methodName)}' does not exist on the wrapped ChatResponseStream.`);
-			return (() => undefined) as any;
+			// Return a typed no-op function that matches the expected signature
+			return ((() => undefined) as unknown as ChatResponseStream[K]);
 		}
-		return ((...args: any[]) => {
+		// Preserve the function signature by casting to the specific method type
+		return ((...args: Parameters<Extract<ChatResponseStream[K], (...args: any[]) => any>>) => {
 			this._codeBlockProcessor.flush();
 			return (method as Function).apply(this._wrapped, args);
-		}) as any;
+		}) as ChatResponseStream[K];
 	}
 
 	button = this.forward('button');

From b4c9a5e7276bd7e15bbf2190e09d8156b8fd028c Mon Sep 17 00:00:00 2001
From: koureasstavros <koureasstavros@gmail.com>
Date: Tue, 4 Nov 2025 12:23:10 +0200
Subject: [PATCH 14/21] Fixed Copilot reviews suggestions

---
 package.json | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/package.json b/package.json
index 22e17f467d..cb84ed3476 100644
--- a/package.json
+++ b/package.json
@@ -3150,6 +3150,12 @@
 									"default": false,
 									"description": "Whether the model supports thinking capabilities"
 								},
+								"temperature": {
+									"type": "number",
+									"description": "Temperature parameter for model responses (typically 0.0 to 2.0)",
+									"minimum": 0,
+									"maximum": 2
+								},
 								"requestHeaders": {
 									"type": "object",
 									"description": "Additional HTTP headers to include with requests to this model. These reserved headers are not allowed and ignored if present: forbidden request headers (https://developer.mozilla.org/en-US/docs/Glossary/Forbidden_request_header), forwarding headers ('forwarded', 'x-forwarded-for', 'x-forwarded-host', 'x-forwarded-proto'), and others ('api-key', 'authorization', 'content-type', 'openai-intent', 'x-github-api-version', 'x-initiator', 'x-interaction-id', 'x-interaction-type', 'x-onbehalf-extension-id', 'x-request-id', 'x-vscode-user-agent-library-version'). Pattern-based forbidden headers ('proxy-*', 'sec-*', 'x-http-method*' with forbidden methods) are also blocked.",

From 9ef329607f468350f07fc39333eb668a239722a9 Mon Sep 17 00:00:00 2001
From: koureasstavros <koureasstavros@gmail.com>
Date: Tue, 4 Nov 2025 13:21:55 +0200
Subject: [PATCH 15/21] Fixed Copilot reviews suggestions

---
 .../byok/vscode-node/azureProvider.ts         | 114 ++++++++----------
 1 file changed, 48 insertions(+), 66 deletions(-)

diff --git a/src/extension/byok/vscode-node/azureProvider.ts b/src/extension/byok/vscode-node/azureProvider.ts
index 71a6fdfed2..c12ffce820 100644
--- a/src/extension/byok/vscode-node/azureProvider.ts
+++ b/src/extension/byok/vscode-node/azureProvider.ts
@@ -19,14 +19,21 @@ export interface AzureUrlOptions {
 }
 
 interface AzureModelConfig {
-	deploymentType?: 'completions' | 'responses';
-	deploymentName?: string;
-	apiVersion?: string;
+	deploymentType: 'completions' | 'responses';
+	deploymentName: string;
+	apiVersion: string;
 	temperature?: number;
-	url: string;
 }
 
-export function resolveAzureUrl(modelId: string, url: string, options?: AzureUrlOptions): string {
+function createAzureUrlOptions(modelId: string, config?: Partial<AzureModelConfig>): AzureUrlOptions {
+	return {
+		deploymentType: config?.deploymentType ?? 'completions',
+		deploymentName: config?.deploymentName ?? modelId,
+		apiVersion: config?.apiVersion ?? '2025-01-01-preview'
+	};
+}
+
+export function resolveAzureUrl(modelId: string, url: string, options: AzureUrlOptions): string {
 
 	// The fully resolved url was already passed in
 	if (hasExplicitApiPath(url)) {
@@ -42,11 +49,7 @@ export function resolveAzureUrl(modelId: string, url: string, options?: AzureUrl
 		url = url.slice(0, -3);
 	}
 
-	// Check if deployment type is specified
-	// If no deployment name is provided, use modelId as fallback
-	const deploymentName = options?.deploymentName || modelId;
-	const deploymentType = options?.deploymentType || 'completions';
-	const apiVersion = options?.apiVersion || '2025-01-01-preview';
+	const { deploymentName, deploymentType, apiVersion } = options;
 
 	// Determine if this is an Azure OpenAI endpoint (requires deployment name in path)
 	const isAzureOpenAIEndpointv1 = url.includes('models.ai.azure.com') || url.includes('inference.ml.azure.com');
@@ -109,65 +112,44 @@ export class AzureBYOKModelProvider extends CustomOAIBYOKModelProvider {
 	}
 
 	protected override resolveUrl(modelId: string, url: string): string {
-		try {
-			// Get model config to access deployment options
-			const modelConfig = this._configurationService?.getConfig(this.getConfigKey()) as Record<string, AzureModelConfig> | undefined;
-
-			const config = modelConfig?.[modelId];
-
-			const options: AzureUrlOptions | undefined = config ? {
-				deploymentType: config.deploymentType || 'completions',
-				deploymentName: config.deploymentName || modelId,
-				apiVersion: config.apiVersion || '2025-01-01-preview'
-			} : undefined;
-
-			const resolvedUrl = resolveAzureUrl(modelId, url, options);
-			return resolvedUrl;
-		} catch (error) {
-			this._logService?.error(`AzureBYOKModelProvider: Error resolving URL for model ${modelId}, falling back to basic resolution:`, error);
-			return resolveAzureUrl(modelId, url, undefined);
-		}
+		// Get model config to access deployment options
+		const modelConfig = this._configurationService?.getConfig(this.getConfigKey()) as Record<string, AzureModelConfig> | undefined;
+		const config = modelConfig?.[modelId];
+		const options = createAzureUrlOptions(modelId, config);
+		return resolveAzureUrl(modelId, url, options);
 	}
 
 	protected override async getModelInfo(modelId: string, apiKey: string | undefined, modelCapabilities?: BYOKModelCapabilities): Promise<IChatModelInformation> {
-		try {
-			// Get model config to check deployment type and deployment name
-			const configKey = this.getConfigKey();
-
-			const modelConfig = this._configurationService?.getConfig(configKey);
-
-			// Safely access the model-specific config
-			let config: AzureModelConfig | undefined;
-			if (modelConfig && typeof modelConfig === 'object' && modelId in modelConfig) {
-				config = (modelConfig as Record<string, AzureModelConfig>)[modelId];
-			}
-
-			const deploymentType = config?.deploymentType || 'completions';
-			// If no deployment name is provided, use modelId as fallback
-			const deploymentName = config?.deploymentName || modelId;
-
-			const modelInfo = await super.getModelInfo(modelId, apiKey, modelCapabilities);
-
-			// Set modelInfo.id to deployment name (or modelId if no deployment name configured)
-			modelInfo.id = deploymentName;
-
-			// Set temperature from config if specified
-			if (config?.temperature !== undefined) {
-				modelInfo.temperature = config.temperature;
-			}
-
-			// Set supported endpoints based on deployment type
-			if (deploymentType === 'responses') {
-				modelInfo.supported_endpoints = [ModelSupportedEndpoint.Responses];
-			} else {
-				// For completions API, only support chat completions
-				modelInfo.supported_endpoints = [ModelSupportedEndpoint.ChatCompletions];
-			}
-
-			return modelInfo;
-		} catch (error) {
-			this._logService?.error(`AzureBYOKModelProvider: Error getting model info for ${modelId}:`, error);
-			throw error;
+		// Get model config to check deployment type and deployment name
+		const configKey = this.getConfigKey();
+
+		const modelConfig = this._configurationService?.getConfig(configKey);
+
+		// Safely access the model-specific config
+		let config: AzureModelConfig | undefined;
+		if (modelConfig && typeof modelConfig === 'object' && modelId in modelConfig) {
+			config = (modelConfig as Record<string, AzureModelConfig>)[modelId];
 		}
+
+		const options = createAzureUrlOptions(modelId, config);
+		const modelInfo = await super.getModelInfo(modelId, apiKey, modelCapabilities);
+
+		// Set modelInfo.id to deployment name (or modelId if no deployment name configured)
+		modelInfo.id = options.deploymentName;
+
+		// Set temperature from config if specified
+		if (config?.temperature !== undefined) {
+			modelInfo.temperature = config.temperature;
+		}
+
+		// Set supported endpoints based on deployment type
+		if (options.deploymentType === 'responses') {
+			modelInfo.supported_endpoints = [ModelSupportedEndpoint.Responses];
+		} else {
+			// For completions API, only support chat completions
+			modelInfo.supported_endpoints = [ModelSupportedEndpoint.ChatCompletions];
+		}
+
+		return modelInfo;
 	}
 }

From 325f4757ba5e97607eff621c89aaab2688042328 Mon Sep 17 00:00:00 2001
From: koureasstavros <koureasstavros@gmail.com>
Date: Tue, 4 Nov 2025 13:31:41 +0200
Subject: [PATCH 16/21] Fixed Copilot reviews suggestions

---
 src/extension/byok/vscode-node/azureProvider.ts | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/src/extension/byok/vscode-node/azureProvider.ts b/src/extension/byok/vscode-node/azureProvider.ts
index c12ffce820..c612ef620a 100644
--- a/src/extension/byok/vscode-node/azureProvider.ts
+++ b/src/extension/byok/vscode-node/azureProvider.ts
@@ -122,8 +122,7 @@ export class AzureBYOKModelProvider extends CustomOAIBYOKModelProvider {
 	protected override async getModelInfo(modelId: string, apiKey: string | undefined, modelCapabilities?: BYOKModelCapabilities): Promise<IChatModelInformation> {
 		// Get model config to check deployment type and deployment name
 		const configKey = this.getConfigKey();
-
-		const modelConfig = this._configurationService?.getConfig(configKey);
+		const modelConfig = this._configurationService.getConfig(configKey);
 
 		// Safely access the model-specific config
 		let config: AzureModelConfig | undefined;

From ce56e2799b845a6dfdc07efc717cb126062ae737 Mon Sep 17 00:00:00 2001
From: koureasstavros <koureasstavros@gmail.com>
Date: Tue, 4 Nov 2025 14:07:12 +0200
Subject: [PATCH 17/21] Fixed Copilot reviews suggestions

---
 src/extension/byok/node/openAIEndpoint.ts          |  7 ++++---
 src/extension/byok/vscode-node/azureProvider.ts    |  3 ++-
 .../codeBlocks/node/codeBlockProcessor.ts          | 14 +++++++++-----
 3 files changed, 15 insertions(+), 9 deletions(-)

diff --git a/src/extension/byok/node/openAIEndpoint.ts b/src/extension/byok/node/openAIEndpoint.ts
index da70c75734..e2d47064e0 100644
--- a/src/extension/byok/node/openAIEndpoint.ts
+++ b/src/extension/byok/node/openAIEndpoint.ts
@@ -300,11 +300,12 @@ export class OpenAIEndpoint extends ChatEndpoint {
 				// Thinking models don't support temperature parameter
 				delete body.temperature;
 				// Thinking models use max_completion_tokens instead of max_tokens
-				body['max_completion_tokens'] = body.max_tokens;
+				if (typeof body.max_tokens !== 'undefined') {
+					body['max_completion_tokens'] = body.max_tokens;
+				}
 			}
 
-			// For thinking models, replace max_tokens with max_completion_tokens
-			// For non-thinking models, removing max_tokens defaults to the maximum which is what we want for BYOK
+			// Removing max_tokens defaults to the maximum which is what we want for BYOK
 			delete body.max_tokens;
 
 			if (!this.useResponsesApi && body.stream) {
diff --git a/src/extension/byok/vscode-node/azureProvider.ts b/src/extension/byok/vscode-node/azureProvider.ts
index c612ef620a..463e5dd4b6 100644
--- a/src/extension/byok/vscode-node/azureProvider.ts
+++ b/src/extension/byok/vscode-node/azureProvider.ts
@@ -133,7 +133,8 @@ export class AzureBYOKModelProvider extends CustomOAIBYOKModelProvider {
 		const options = createAzureUrlOptions(modelId, config);
 		const modelInfo = await super.getModelInfo(modelId, apiKey, modelCapabilities);
 
-		// Set modelInfo.id to deployment name (or modelId if no deployment name configured)
+		// IMPORTANT: Override modelInfo.id with the deployment name (handled above).
+		// This is required because Azure OpenAI uses the deployment name for API routing.
 		modelInfo.id = options.deploymentName;
 
 		// Set temperature from config if specified
diff --git a/src/extension/codeBlocks/node/codeBlockProcessor.ts b/src/extension/codeBlocks/node/codeBlockProcessor.ts
index 724d5fba10..c23f7ef92f 100644
--- a/src/extension/codeBlocks/node/codeBlockProcessor.ts
+++ b/src/extension/codeBlocks/node/codeBlockProcessor.ts
@@ -108,14 +108,18 @@ export class CodeBlockTrackingChatResponseStream implements ChatResponseStream {
 		const method = this._wrapped[methodName];
 		if (typeof method !== 'function') {
 			this._logService.warn(`[CodeBlockTrackingChatResponseStream] Method '${String(methodName)}' does not exist on the wrapped ChatResponseStream.`);
-			// Return a typed no-op function that matches the expected signature
-			return ((() => undefined) as unknown as ChatResponseStream[K]);
+			// Create a proper no-op function that matches the method signature
+			const noOp = (..._args: any[]): any => {
+				return undefined;
+			};
+			return noOp as ChatResponseStream[K];
 		}
-		// Preserve the function signature by casting to the specific method type
-		return ((...args: Parameters<Extract<ChatResponseStream[K], (...args: any[]) => any>>) => {
+		// Preserve the function signature without double casting
+		const boundMethod = (...args: any[]) => {
 			this._codeBlockProcessor.flush();
 			return (method as Function).apply(this._wrapped, args);
-		}) as ChatResponseStream[K];
+		};
+		return boundMethod as ChatResponseStream[K];
 	}
 
 	button = this.forward('button');

From 5f433070cfdf2dec060bb4399cdacbf678f73516 Mon Sep 17 00:00:00 2001
From: koureasstavros <koureasstavros@gmail.com>
Date: Fri, 7 Nov 2025 10:41:29 +0200
Subject: [PATCH 18/21] Fixed Copilot reviews suggestions

---
 .../azure-provider-settings.instructions.md   | 108 +++---------------
 package.json                                  |  24 ++--
 src/extension/byok/node/openAIEndpoint.ts     |  10 +-
 .../byok/vscode-node/azureProvider.ts         |   2 +-
 4 files changed, 34 insertions(+), 110 deletions(-)

diff --git a/.github/instructions/azure-provider-settings.instructions.md b/.github/instructions/azure-provider-settings.instructions.md
index afd840a021..e4fb9c666a 100644
--- a/.github/instructions/azure-provider-settings.instructions.md
+++ b/.github/instructions/azure-provider-settings.instructions.md
@@ -22,13 +22,12 @@ The `chat.azureModels` setting accepts a record of model configurations where ea
       "deploymentType": "completions" | "responses",  // API type (default: "completions")
       "deploymentName": string,          // Azure deployment name (optional, defaults to modelId)
       "apiVersion": string,              // Azure API version (optional, default: "2025-01-01-preview")
-      "temperature": number,             // Temperature setting (optional, 0.0-2.0)
-      "toolCalling": boolean,            // Whether model supports function/tool calling
-      "vision": boolean,                 // Whether model supports vision/image inputs
       "maxInputTokens": number,          // Maximum input context window size
       "maxOutputTokens": number,         // Maximum output tokens per response
-      "requiresAPIKey": boolean,         // Whether model requires API key (optional, default: true)
+      "toolCalling": boolean,            // Whether model supports function/tool calling
+      "temperature": number,             // Temperature setting (optional, 0.0-2.0)
       "thinking": boolean                // Whether model supports thinking/reasoning tokens (optional)
+      "vision": boolean,                 // Whether model supports vision/image inputs (optional)
     }
   }
 }
@@ -69,10 +68,12 @@ The `url` field should contain the **base Azure endpoint URL** without the API p
       "deploymentType": "completions",
       "deploymentName": "gpt-4-turbo-deployment",
       "apiVersion": "2024-08-01-preview",
-      "toolCalling": true,
-      "vision": false,
       "maxInputTokens": 128000,
-      "maxOutputTokens": 4096
+      "maxOutputTokens": 4096,
+      "toolCalling": true,
+      "temperature": 0.7,
+      "thinking": false,
+      "vision": false
     }
   }
 }
@@ -98,11 +99,12 @@ Uses the Azure OpenAI Responses API for streaming responses with structured outp
       "deploymentType": "responses",
       "deploymentName": "o1-preview-deployment",
       "apiVersion": "2025-01-01-preview",
+      "maxInputTokens": 128000,
+      "maxOutputTokens": 32768,
       "toolCalling": false,
-      "vision": true,
+      "temperature": 1.0,
       "thinking": true,
-      "maxInputTokens": 128000,
-      "maxOutputTokens": 32768
+      "vision": true
     }
   }
 }
@@ -115,89 +117,6 @@ Uses the Azure OpenAI Responses API for streaming responses with structured outp
 - **For Completions API**: Deployment name is included in the URL path
 - **For Responses API**: Deployment name is sent as the `model` field in the request body
 
-## Complete Configuration Examples
-
-### Example 1: Azure OpenAI - Chat Completions
-```json
-{
-  "chat.azureModels": {
-    "gpt-4": {
-      "name": "GPT-4",
-      "url": "https://my-eastus-resource.openai.azure.com",
-      "deploymentType": "completions",
-      "deploymentName": "gpt-4-prod",
-      "apiVersion": "2024-08-01-preview",
-      "temperature": 0.7,
-      "toolCalling": true,
-      "vision": false,
-      "maxInputTokens": 8192,
-      "maxOutputTokens": 4096,
-      "requiresAPIKey": true
-    }
-  }
-}
-```
-
-### Example 2: Azure AI - Responses API with Reasoning
-```json
-{
-  "chat.azureModels": {
-    "o1": {
-      "name": "OpenAI o1",
-      "url": "https://my-resource.models.ai.azure.com",
-      "deploymentType": "responses",
-      "deploymentName": "o1-deployment",
-      "apiVersion": "2025-01-01-preview",
-      "toolCalling": false,
-      "vision": true,
-      "thinking": true,
-      "maxInputTokens": 200000,
-      "maxOutputTokens": 100000,
-      "requiresAPIKey": true
-    }
-  }
-}
-```
-
-### Example 3: Multiple Models
-```json
-{
-  "chat.azureModels": {
-    "gpt-35-turbo": {
-      "name": "GPT-3.5 Turbo",
-      "url": "https://my-resource.openai.azure.com",
-      "deploymentType": "completions",
-      "deploymentName": "gpt-35-turbo-16k",
-      "toolCalling": true,
-      "vision": false,
-      "maxInputTokens": 16384,
-      "maxOutputTokens": 4096
-    },
-    "gpt-4-vision": {
-      "name": "GPT-4 Vision",
-      "url": "https://my-resource.openai.azure.com",
-      "deploymentType": "completions",
-      "deploymentName": "gpt-4-vision-preview",
-      "toolCalling": true,
-      "vision": true,
-      "maxInputTokens": 128000,
-      "maxOutputTokens": 4096
-    },
-    "o1-preview": {
-      "name": "OpenAI o1 Preview",
-      "url": "https://my-resource.openai.azure.com",
-      "deploymentType": "responses",
-      "deploymentName": "o1-preview",
-      "toolCalling": false,
-      "vision": true,
-      "thinking": true,
-      "maxInputTokens": 128000,
-      "maxOutputTokens": 32768
-    }
-  }
-}
-```
-
 ## Implementation Details
 
 ### URL Resolution Logic (`resolveAzureUrl`)
@@ -239,7 +158,7 @@ The `getModelInfo` method in `AzureBYOKModelProvider`:
 1. **Use explicit deployment names**: Always specify `deploymentName` to avoid confusion with model IDs
 2. **Match API versions**: Use the API version that matches your Azure deployment capabilities
 3. **Set accurate token limits**: Configure `maxInputTokens` and `maxOutputTokens` based on your deployment
-4. **Enable appropriate capabilities**: Set `toolCalling`, `vision`, and `thinking` flags based on model support
+4. **Enable appropriate capabilities**: Set `toolCalling`, `thinking` and `vision`, flags based on model support
 5. **Test endpoint URLs**: Verify base URLs are correct and accessible before adding models
 6. **Group related models**: Use descriptive model IDs for easy identification in the UI
 7. **Document custom configurations**: Add comments in settings.json to explain non-standard configurations
@@ -249,7 +168,6 @@ The `getModelInfo` method in `AzureBYOKModelProvider`:
 ### Common Issues
 
 **Issue**: Model not appearing in selection
-- **Check**: Verify `requiresAPIKey` is set correctly and API key is configured
 - **Check**: Ensure all required fields are present in configuration
 
 **Issue**: 404 errors when using model
diff --git a/package.json b/package.json
index cb84ed3476..fb2b6692f1 100644
--- a/package.json
+++ b/package.json
@@ -3059,20 +3059,20 @@
 									"type": "boolean",
 									"description": "Whether the model supports tool calling"
 								},
-								"vision": {
-									"type": "boolean",
-									"description": "Whether the model supports vision capabilities"
+								"temperature": {
+									"type": "number",
+									"description": "Temperature parameter for model responses (typically 0.0 to 2.0)",
+									"minimum": 0,
+									"maximum": 2
 								},
 								"thinking": {
 									"type": "boolean",
 									"default": false,
 									"description": "Whether the model supports thinking capabilities"
 								},
-								"temperature": {
-									"type": "number",
-									"description": "Temperature parameter for model responses (typically 0.0 to 2.0)",
-									"minimum": 0,
-									"maximum": 2
+								"vision": {
+									"type": "boolean",
+									"description": "Whether the model supports vision capabilities"
 								},
 								"requestHeaders": {
 									"type": "object",
@@ -3085,10 +3085,12 @@
 							"required": [
 								"name",
 								"url",
-								"toolCalling",
-								"vision",
+								"deploymentName",
+								"deploymentType",
 								"maxInputTokens",
-								"maxOutputTokens"
+								"maxOutputTokens",
+								"toolCalling",
+								"vision"
 							],
 							"additionalProperties": false
 						},
diff --git a/src/extension/byok/node/openAIEndpoint.ts b/src/extension/byok/node/openAIEndpoint.ts
index e2d47064e0..42b34572ea 100644
--- a/src/extension/byok/node/openAIEndpoint.ts
+++ b/src/extension/byok/node/openAIEndpoint.ts
@@ -302,11 +302,15 @@ export class OpenAIEndpoint extends ChatEndpoint {
 				// Thinking models use max_completion_tokens instead of max_tokens
 				if (typeof body.max_tokens !== 'undefined') {
 					body['max_completion_tokens'] = body.max_tokens;
+					delete body.max_tokens;
+				}
+			}
+			else {
+				if (typeof body.max_tokens !== 'undefined') {
+					body['max_completion_tokens'] = body.max_tokens;
+					delete body.max_tokens;
 				}
 			}
-
-			// Removing max_tokens defaults to the maximum which is what we want for BYOK
-			delete body.max_tokens;
 
 			if (!this.useResponsesApi && body.stream) {
 				body['stream_options'] = { 'include_usage': true };
diff --git a/src/extension/byok/vscode-node/azureProvider.ts b/src/extension/byok/vscode-node/azureProvider.ts
index 463e5dd4b6..52074255c1 100644
--- a/src/extension/byok/vscode-node/azureProvider.ts
+++ b/src/extension/byok/vscode-node/azureProvider.ts
@@ -133,7 +133,7 @@ export class AzureBYOKModelProvider extends CustomOAIBYOKModelProvider {
 		const options = createAzureUrlOptions(modelId, config);
 		const modelInfo = await super.getModelInfo(modelId, apiKey, modelCapabilities);
 
-		// IMPORTANT: Override modelInfo.id with the deployment name (handled above).
+		// Override modelInfo.id with the deployment name (deployment name is also handled).
 		// This is required because Azure OpenAI uses the deployment name for API routing.
 		modelInfo.id = options.deploymentName;
 

From 3307f20ba882264ede966ee417ac4c42cebef837 Mon Sep 17 00:00:00 2001
From: koureasstavros <koureasstavros@gmail.com>
Date: Fri, 7 Nov 2025 10:57:57 +0200
Subject: [PATCH 19/21] Fixed Copilot reviews suggestions

---
 .../instructions/azure-provider-settings.instructions.md    | 6 +++---
 package.json                                                | 2 --
 src/extension/byok/node/openAIEndpoint.ts                   | 1 -
 3 files changed, 3 insertions(+), 6 deletions(-)

diff --git a/.github/instructions/azure-provider-settings.instructions.md b/.github/instructions/azure-provider-settings.instructions.md
index e4fb9c666a..b28de48162 100644
--- a/.github/instructions/azure-provider-settings.instructions.md
+++ b/.github/instructions/azure-provider-settings.instructions.md
@@ -19,14 +19,14 @@ The `chat.azureModels` setting accepts a record of model configurations where ea
     "[modelId]": {
       "name": string,                    // Display name for the model
       "url": string,                     // Azure endpoint URL (base URL without API path)
-      "deploymentType": "completions" | "responses",  // API type (default: "completions")
+      "deploymentType": "completions" | "responses",  // API type (optional, defaults to "completions")
       "deploymentName": string,          // Azure deployment name (optional, defaults to modelId)
       "apiVersion": string,              // Azure API version (optional, default: "2025-01-01-preview")
       "maxInputTokens": number,          // Maximum input context window size
       "maxOutputTokens": number,         // Maximum output tokens per response
       "toolCalling": boolean,            // Whether model supports function/tool calling
       "temperature": number,             // Temperature setting (optional, 0.0-2.0)
-      "thinking": boolean                // Whether model supports thinking/reasoning tokens (optional)
+      "thinking": boolean,               // Whether model supports thinking/reasoning tokens (optional)
       "vision": boolean,                 // Whether model supports vision/image inputs (optional)
     }
   }
@@ -158,7 +158,7 @@ The `getModelInfo` method in `AzureBYOKModelProvider`:
 1. **Use explicit deployment names**: Always specify `deploymentName` to avoid confusion with model IDs
 2. **Match API versions**: Use the API version that matches your Azure deployment capabilities
 3. **Set accurate token limits**: Configure `maxInputTokens` and `maxOutputTokens` based on your deployment
-4. **Enable appropriate capabilities**: Set `toolCalling`, `thinking` and `vision`, flags based on model support
+4. **Enable appropriate capabilities**: Set `toolCalling`, `thinking` and `vision` flags based on model support
 5. **Test endpoint URLs**: Verify base URLs are correct and accessible before adding models
 6. **Group related models**: Use descriptive model IDs for easy identification in the UI
 7. **Document custom configurations**: Add comments in settings.json to explain non-standard configurations
diff --git a/package.json b/package.json
index fb2b6692f1..f35e2320c9 100644
--- a/package.json
+++ b/package.json
@@ -3085,8 +3085,6 @@
 							"required": [
 								"name",
 								"url",
-								"deploymentName",
-								"deploymentType",
 								"maxInputTokens",
 								"maxOutputTokens",
 								"toolCalling",
diff --git a/src/extension/byok/node/openAIEndpoint.ts b/src/extension/byok/node/openAIEndpoint.ts
index 42b34572ea..0e89f2eb17 100644
--- a/src/extension/byok/node/openAIEndpoint.ts
+++ b/src/extension/byok/node/openAIEndpoint.ts
@@ -307,7 +307,6 @@ export class OpenAIEndpoint extends ChatEndpoint {
 			}
 			else {
 				if (typeof body.max_tokens !== 'undefined') {
-					body['max_completion_tokens'] = body.max_tokens;
 					delete body.max_tokens;
 				}
 			}

From c37f2c962893b8bc7e4360408bbbc0c2fe23660f Mon Sep 17 00:00:00 2001
From: koureasstavros <koureasstavros@gmail.com>
Date: Fri, 7 Nov 2025 11:34:29 +0200
Subject: [PATCH 20/21] Fixed Copilot reviews suggestions

---
 .../azure-provider-settings.instructions.md   | 12 ++++++-----
 package.json                                  |  6 +++++-
 src/extension/byok/node/openAIEndpoint.ts     |  1 +
 .../byok/vscode-node/azureProvider.ts         | 20 +++++++++++--------
 .../byok/vscode-node/customOAIProvider.ts     |  7 ++-----
 5 files changed, 27 insertions(+), 19 deletions(-)

diff --git a/.github/instructions/azure-provider-settings.instructions.md b/.github/instructions/azure-provider-settings.instructions.md
index b28de48162..13bf2165af 100644
--- a/.github/instructions/azure-provider-settings.instructions.md
+++ b/.github/instructions/azure-provider-settings.instructions.md
@@ -19,15 +19,17 @@ The `chat.azureModels` setting accepts a record of model configurations where ea
     "[modelId]": {
       "name": string,                    // Display name for the model
       "url": string,                     // Azure endpoint URL (base URL without API path)
-      "deploymentType": "completions" | "responses",  // API type (optional, defaults to "completions")
-      "deploymentName": string,          // Azure deployment name (optional, defaults to modelId)
-      "apiVersion": string,              // Azure API version (optional, default: "2025-01-01-preview")
+      "deploymentType"?: "completions" | "responses",  // API type (optional, defaults to "completions")
+      "deploymentName"?: string,          // Azure deployment name (optional, defaults to modelId)
+      "apiVersion"?: string,              // Azure API version (optional, default: "2025-01-01-preview")
       "maxInputTokens": number,          // Maximum input context window size
       "maxOutputTokens": number,         // Maximum output tokens per response
       "toolCalling": boolean,            // Whether model supports function/tool calling
-      "temperature": number,             // Temperature setting (optional, 0.0-2.0)
-      "thinking": boolean,               // Whether model supports thinking/reasoning tokens (optional)
+      "temperature"?: number,             // Temperature setting (optional, 0.0-2.0)
+      "thinking"?: boolean,               // Whether model supports thinking/reasoning tokens (optional)
       "vision": boolean,                 // Whether model supports vision/image inputs (optional)
+      "requestHeaders"?: Record<string, string>,  // Custom HTTP headers for requests (optional)
+      "requiresAPIKey"?: boolean,        // Whether model requires API key (optional, default: true)
     }
   }
 }
diff --git a/package.json b/package.json
index f35e2320c9..7d0bb66d7b 100644
--- a/package.json
+++ b/package.json
@@ -3040,7 +3040,11 @@
 									"description": "Azure deployment name for the model"
 								},
 								"deploymentType": {
-									"type": "string",
+									"enum": [
+										"completions",
+										"responses"
+									],
+									"default": "completions",
 									"description": "Type of Azure deployment"
 								},
 								"apiVersion": {
diff --git a/src/extension/byok/node/openAIEndpoint.ts b/src/extension/byok/node/openAIEndpoint.ts
index 0e89f2eb17..af26953ac8 100644
--- a/src/extension/byok/node/openAIEndpoint.ts
+++ b/src/extension/byok/node/openAIEndpoint.ts
@@ -306,6 +306,7 @@ export class OpenAIEndpoint extends ChatEndpoint {
 				}
 			}
 			else {
+				// Non thinking models does not use max_tokens
 				if (typeof body.max_tokens !== 'undefined') {
 					delete body.max_tokens;
 				}
diff --git a/src/extension/byok/vscode-node/azureProvider.ts b/src/extension/byok/vscode-node/azureProvider.ts
index 52074255c1..b7649c3b35 100644
--- a/src/extension/byok/vscode-node/azureProvider.ts
+++ b/src/extension/byok/vscode-node/azureProvider.ts
@@ -19,10 +19,19 @@ export interface AzureUrlOptions {
 }
 
 interface AzureModelConfig {
-	deploymentType: 'completions' | 'responses';
-	deploymentName: string;
-	apiVersion: string;
+	name: string;
+	url: string;
+	deploymentType?: 'completions' | 'responses';
+	deploymentName?: string;
+	apiVersion?: string;
+	maxInputTokens: number;
+	maxOutputTokens: number;
+	toolCalling: boolean;
+	thinking?: boolean;
+	vision: boolean;
 	temperature?: number;
+	requiresAPIKey?: boolean;
+	requestHeaders?: Record<string, string>;
 }
 
 function createAzureUrlOptions(modelId: string, config?: Partial<AzureModelConfig>): AzureUrlOptions {
@@ -137,11 +146,6 @@ export class AzureBYOKModelProvider extends CustomOAIBYOKModelProvider {
 		// This is required because Azure OpenAI uses the deployment name for API routing.
 		modelInfo.id = options.deploymentName;
 
-		// Set temperature from config if specified
-		if (config?.temperature !== undefined) {
-			modelInfo.temperature = config.temperature;
-		}
-
 		// Set supported endpoints based on deployment type
 		if (options.deploymentType === 'responses') {
 			modelInfo.supported_endpoints = [ModelSupportedEndpoint.Responses];
diff --git a/src/extension/byok/vscode-node/customOAIProvider.ts b/src/extension/byok/vscode-node/customOAIProvider.ts
index 7ac98615b7..e75b4adec1 100644
--- a/src/extension/byok/vscode-node/customOAIProvider.ts
+++ b/src/extension/byok/vscode-node/customOAIProvider.ts
@@ -47,17 +47,14 @@ export function hasExplicitApiPath(url: string): boolean {
 interface UserModelConfig {
 	name: string;
 	url: string;
-	deploymentType?: 'completions' | 'responses';
-	deploymentName?: string;
-	apiVersion?: string;
 	maxInputTokens: number;
 	maxOutputTokens: number;
 	requiresAPIKey: boolean;
 	toolCalling: boolean;
 	editTools?: EndpointEditToolName[];
-	vision: boolean;
-	thinking?: boolean;
 	temperature?: number;
+	thinking?: boolean;
+	vision: boolean;
 	requestHeaders?: Record<string, string>;
 }
 

From 760a99a58eff9c83473a553b77646b9c4fbff788 Mon Sep 17 00:00:00 2001
From: koureasstavros <koureasstavros@gmail.com>
Date: Fri, 7 Nov 2025 12:01:49 +0200
Subject: [PATCH 21/21] Fixed Copilot reviews suggestions

---
 .../azure-provider-settings.instructions.md   | 21 ++++---
 package.json                                  | 61 +++++++++++--------
 src/extension/byok/node/openAIEndpoint.ts     |  2 +-
 .../byok/vscode-node/azureProvider.ts         |  7 ++-
 .../byok/vscode-node/customOAIProvider.ts     | 10 +--
 5 files changed, 56 insertions(+), 45 deletions(-)

diff --git a/.github/instructions/azure-provider-settings.instructions.md b/.github/instructions/azure-provider-settings.instructions.md
index 13bf2165af..3262d889c5 100644
--- a/.github/instructions/azure-provider-settings.instructions.md
+++ b/.github/instructions/azure-provider-settings.instructions.md
@@ -17,19 +17,20 @@ The `chat.azureModels` setting accepts a record of model configurations where ea
 {
   "chat.azureModels": {
     "[modelId]": {
-      "name": string,                    // Display name for the model
-      "url": string,                     // Azure endpoint URL (base URL without API path)
+      "name": string,                     // Display name for the model
+      "url": string,                      // Azure endpoint URL (base URL without API path)
       "deploymentType"?: "completions" | "responses",  // API type (optional, defaults to "completions")
       "deploymentName"?: string,          // Azure deployment name (optional, defaults to modelId)
       "apiVersion"?: string,              // Azure API version (optional, default: "2025-01-01-preview")
-      "maxInputTokens": number,          // Maximum input context window size
-      "maxOutputTokens": number,         // Maximum output tokens per response
-      "toolCalling": boolean,            // Whether model supports function/tool calling
-      "temperature"?: number,             // Temperature setting (optional, 0.0-2.0)
-      "thinking"?: boolean,               // Whether model supports thinking/reasoning tokens (optional)
-      "vision": boolean,                 // Whether model supports vision/image inputs (optional)
-      "requestHeaders"?: Record<string, string>,  // Custom HTTP headers for requests (optional)
-      "requiresAPIKey"?: boolean,        // Whether model requires API key (optional, default: true)
+      "maxInputTokens": number,           // Maximum input context window size
+      "maxOutputTokens": number,          // Maximum output tokens per response
+      "toolCalling"?: boolean,            // Whether model supports function/tool calling (optional, default: false)
+      "editTools"?: string[],             // List of supported edit tools (optional, default: [])
+      "temperature"?: number,             // Temperature setting (optional, 0.0-2.0, default: 1.0)
+      "thinking"?: boolean,               // Whether model supports thinking/reasoning tokens (optional, default: false)
+      "vision"?: boolean,                 // Whether model supports vision/image inputs (optional, default: false)
+      "requestHeaders"?: Record<string, string>,  // Custom HTTP headers for requests (optional, default: null)
+      "requiresAPIKey"?: boolean,         // Whether model requires API key (optional, default: true)
     }
   }
 }
diff --git a/package.json b/package.json
index 7d0bb66d7b..b916d5df52 100644
--- a/package.json
+++ b/package.json
@@ -3063,6 +3063,19 @@
 									"type": "boolean",
 									"description": "Whether the model supports tool calling"
 								},
+								"editTools": {
+									"type": "array",
+									"description": "List of edit tools supported by the model. If this is not configured, the editor will try multiple edit tools and pick the best one.\n\n- 'find-replace': Find and replace text in a document.\n- 'multi-find-replace': Find and replace text in a document.\n- 'apply-patch': A file-oriented diff format used by some OpenAI models\n- 'code-rewrite': A general but slower editing tool that allows the model to rewrite and code snippet and provide only the replacement to the editor.",
+									"items": {
+										"type": "string",
+										"enum": [
+											"find-replace",
+											"multi-find-replace",
+											"apply-patch",
+											"code-rewrite"
+										]
+									}
+								},
 								"temperature": {
 									"type": "number",
 									"description": "Temperature parameter for model responses (typically 0.0 to 2.0)",
@@ -3090,9 +3103,7 @@
 								"name",
 								"url",
 								"maxInputTokens",
-								"maxOutputTokens",
-								"toolCalling",
-								"vision"
+								"maxOutputTokens"
 							],
 							"additionalProperties": false
 						},
@@ -3119,23 +3130,6 @@
 									"type": "boolean",
 									"description": "Whether the model supports tool calling"
 								},
-								"vision": {
-									"type": "boolean",
-									"description": "Whether the model supports vision capabilities"
-								},
-								"maxInputTokens": {
-									"type": "number",
-									"description": "Maximum number of input tokens supported by the model"
-								},
-								"maxOutputTokens": {
-									"type": "number",
-									"description": "Maximum number of output tokens supported by the model"
-								},
-								"requiresAPIKey": {
-									"type": "boolean",
-									"description": "Whether the model requires an API key for authentication",
-									"default": true
-								},
 								"editTools": {
 									"type": "array",
 									"description": "List of edit tools supported by the model. If this is not configured, the editor will try multiple edit tools and pick the best one.\n\n- 'find-replace': Find and replace text in a document.\n- 'multi-find-replace': Find and replace text in a document.\n- 'apply-patch': A file-oriented diff format used by some OpenAI models\n- 'code-rewrite': A general but slower editing tool that allows the model to rewrite and code snippet and provide only the replacement to the editor.",
@@ -3149,16 +3143,28 @@
 										]
 									}
 								},
+								"temperature": {
+									"type": "number",
+									"description": "Temperature parameter for model responses (typically 0.0 to 2.0)",
+									"minimum": 0,
+									"maximum": 2
+								},
 								"thinking": {
 									"type": "boolean",
 									"default": false,
 									"description": "Whether the model supports thinking capabilities"
 								},
-								"temperature": {
+								"vision": {
+									"type": "boolean",
+									"description": "Whether the model supports vision capabilities"
+								},
+								"maxInputTokens": {
 									"type": "number",
-									"description": "Temperature parameter for model responses (typically 0.0 to 2.0)",
-									"minimum": 0,
-									"maximum": 2
+									"description": "Maximum number of input tokens supported by the model"
+								},
+								"maxOutputTokens": {
+									"type": "number",
+									"description": "Maximum number of output tokens supported by the model"
 								},
 								"requestHeaders": {
 									"type": "object",
@@ -3166,13 +3172,16 @@
 									"additionalProperties": {
 										"type": "string"
 									}
+								},
+								"requiresAPIKey": {
+									"type": "boolean",
+									"description": "Whether the model requires an API key for authentication",
+									"default": true
 								}
 							},
 							"required": [
 								"name",
 								"url",
-								"toolCalling",
-								"vision",
 								"maxInputTokens",
 								"maxOutputTokens",
 								"requiresAPIKey"
diff --git a/src/extension/byok/node/openAIEndpoint.ts b/src/extension/byok/node/openAIEndpoint.ts
index af26953ac8..7a146de833 100644
--- a/src/extension/byok/node/openAIEndpoint.ts
+++ b/src/extension/byok/node/openAIEndpoint.ts
@@ -306,7 +306,7 @@ export class OpenAIEndpoint extends ChatEndpoint {
 				}
 			}
 			else {
-				// Non thinking models does not use max_tokens
+				// For non-thinking models, removing max_tokens defaults to the maximum allowed value, which is what we want for BYOK
 				if (typeof body.max_tokens !== 'undefined') {
 					delete body.max_tokens;
 				}
diff --git a/src/extension/byok/vscode-node/azureProvider.ts b/src/extension/byok/vscode-node/azureProvider.ts
index b7649c3b35..0e6b084c07 100644
--- a/src/extension/byok/vscode-node/azureProvider.ts
+++ b/src/extension/byok/vscode-node/azureProvider.ts
@@ -4,7 +4,7 @@
  *--------------------------------------------------------------------------------------------*/
 
 import { ConfigKey, IConfigurationService } from '../../../platform/configuration/common/configurationService';
-import { IChatModelInformation, ModelSupportedEndpoint } from '../../../platform/endpoint/common/endpointProvider';
+import { EndpointEditToolName, IChatModelInformation, ModelSupportedEndpoint } from '../../../platform/endpoint/common/endpointProvider';
 import { ILogService } from '../../../platform/log/common/logService';
 import { IExperimentationService } from '../../../platform/telemetry/common/nullExperimentationService';
 import { IInstantiationService } from '../../../util/vs/platform/instantiation/common/instantiation';
@@ -26,9 +26,10 @@ interface AzureModelConfig {
 	apiVersion?: string;
 	maxInputTokens: number;
 	maxOutputTokens: number;
-	toolCalling: boolean;
+	toolCalling?: boolean;
+	editTools?: EndpointEditToolName[];
 	thinking?: boolean;
-	vision: boolean;
+	vision?: boolean;
 	temperature?: number;
 	requiresAPIKey?: boolean;
 	requestHeaders?: Record<string, string>;
diff --git a/src/extension/byok/vscode-node/customOAIProvider.ts b/src/extension/byok/vscode-node/customOAIProvider.ts
index e75b4adec1..3e2e5876a7 100644
--- a/src/extension/byok/vscode-node/customOAIProvider.ts
+++ b/src/extension/byok/vscode-node/customOAIProvider.ts
@@ -49,13 +49,13 @@ interface UserModelConfig {
 	url: string;
 	maxInputTokens: number;
 	maxOutputTokens: number;
-	requiresAPIKey: boolean;
-	toolCalling: boolean;
+	toolCalling?: boolean;
 	editTools?: EndpointEditToolName[];
 	temperature?: number;
 	thinking?: boolean;
-	vision: boolean;
+	vision?: boolean;
 	requestHeaders?: Record<string, string>;
+	requiresAPIKey: boolean;
 }
 
 interface CustomOAIModelInfo extends LanguageModelChatInformation {
@@ -124,9 +124,9 @@ export class CustomOAIBYOKModelProvider implements BYOKModelProvider<CustomOAIMo
 				url: resolvedUrl,
 				maxInputTokens: modelInfo.maxInputTokens,
 				maxOutputTokens: modelInfo.maxOutputTokens,
-				toolCalling: modelInfo.toolCalling,
+				toolCalling: modelInfo.toolCalling ?? false,
 				editTools: modelInfo.editTools,
-				vision: modelInfo.vision,
+				vision: modelInfo.vision ?? false,
 				thinking: modelInfo.thinking,
 				temperature: modelInfo.temperature,
 				requestHeaders: modelInfo.requestHeaders ? { ...modelInfo.requestHeaders } : undefined,