diff --git a/src/api/providers/fetchers/modelCache.ts b/src/api/providers/fetchers/modelCache.ts
index 860ca5155c..77855fd435 100644
--- a/src/api/providers/fetchers/modelCache.ts
+++ b/src/api/providers/fetchers/modelCache.ts
@@ -21,7 +21,7 @@ import { getVercelAiGatewayModels } from "./vercel-ai-gateway"
 import { getRequestyModels } from "./requesty"
 import { getLiteLLMModels } from "./litellm"
 import { GetModelsOptions } from "../../../shared/api"
-import { getOllamaModels } from "./ollama"
+import { discoverOllamaModelsWithSorting } from "./ollama"
 import { getLMStudioModels } from "./lmstudio"
 import { getRooModels } from "./roo"
 
@@ -72,14 +72,21 @@ async function fetchModelsFromProvider(options: GetModelsOptions): Promise<Model
 			// Type safety ensures apiKey and baseUrl are always provided for LiteLLM.
 			models = await getLiteLLMModels(options.apiKey, options.baseUrl)
 			break
-		case "ollama":
-			models = await getOllamaModels(options.baseUrl, options.apiKey, {
+		case "ollama": {
+			// Use discoverOllamaModelsWithSorting as the single discovery path,
+			// extracting only the tools-supporting models for the model cache.
+			const discoveryResult = await discoverOllamaModelsWithSorting(options.baseUrl, options.apiKey, {
 				modelDiscoveryTimeout: options.ollamaModelDiscoveryTimeout,
 				maxRetries: options.ollamaMaxRetries,
 				retryDelay: options.ollamaRetryDelay,
 				enableLogging: options.ollamaEnableLogging,
 			})
+			models = {}
+			for (const m of discoveryResult.modelsWithTools) {
+				models[m.name] = m.modelInfo
+			}
 			break
+		}
 		case "lmstudio":
 			models = await getLMStudioModels(options.baseUrl)
 			break
diff --git a/src/api/providers/fetchers/ollama.ts b/src/api/providers/fetchers/ollama.ts
index bc57b25ee4..0f19807915 100644
--- a/src/api/providers/fetchers/ollama.ts
+++ b/src/api/providers/fetchers/ollama.ts
@@ -25,6 +25,15 @@ interface OllamaAxiosConfig {
 	enableLogging?: boolean
 }
 
+/**
+ * Extended Axios request config with retry count and timing metadata.
+ * Used by retry and logging interceptors to track state across requests.
+ */
+interface OllamaInternalAxiosConfig extends InternalAxiosRequestConfig {
+	__retryCount?: number
+	metadata?: { startTime: number }
+}
+
 export function createOllamaAxiosInstance(config: OllamaAxiosConfig = {}): AxiosInstance {
 	const {
 		baseUrl = "http://localhost:11434",
@@ -64,7 +73,10 @@ function setupRetryInterceptor(instance: AxiosInstance, config: { retries: numbe
 	instance.interceptors.response.use(
 		(response) => response,
 		async (error: AxiosError) => {
-			const axiosConfig = error.config as any
+			const axiosConfig = error.config as OllamaInternalAxiosConfig | undefined
+			if (!axiosConfig) {
+				return Promise.reject(error)
+			}
 
 			axiosConfig.__retryCount = axiosConfig.__retryCount || 0
 			if (axiosConfig.__retryCount >= config.retries) {
@@ -94,7 +106,7 @@ function setupRetryInterceptor(instance: AxiosInstance, config: { retries: numbe
 
 function setupLoggingInterceptor(instance: AxiosInstance) {
 	instance.interceptors.request.use((config: InternalAxiosRequestConfig) => {
-		;(config as any).metadata = { startTime: Date.now() }
+		;(config as OllamaInternalAxiosConfig).metadata = { startTime: Date.now() }
 		console.debug("[Ollama] Request:", {
 			method: config.method?.toUpperCase(),
 			url: `${config.baseURL}${config.url}`,
@@ -106,7 +118,7 @@ function setupLoggingInterceptor(instance: AxiosInstance) {
 
 	instance.interceptors.response.use(
 		(response: AxiosResponse) => {
-			const startTime = (response.config as any).metadata?.startTime
+			const startTime = (response.config as OllamaInternalAxiosConfig).metadata?.startTime
 			const duration = startTime ? Date.now() - startTime : undefined
 			console.debug("[Ollama] Response:", {
 				status: response.status,
@@ -118,7 +130,7 @@ function setupLoggingInterceptor(instance: AxiosInstance) {
 			return response
 		},
 		(error: AxiosError) => {
-			const startTime = (error.config as any)?.metadata?.startTime
+			const startTime = (error.config as OllamaInternalAxiosConfig | undefined)?.metadata?.startTime
 			const duration = startTime ? Date.now() - startTime : undefined
 			console.error("[Ollama] Error:", {
 				code: error.code,
@@ -361,7 +373,7 @@ export async function discoverOllamaModelsWithSorting(
 					const parsedDetail = OllamaModelInfoResponseSchema.safeParse(detailResponse.data)
 					if (!parsedDetail.success) {
 						// If validation fails, check if required fields exist in raw data
-						const rawData = detailResponse.data as any
+						const rawData = detailResponse.data as Partial<OllamaModelInfoResponse>
 						if (!rawData?.details?.family || !rawData?.details?.parameter_size) {
 							if (config?.enableLogging) {
 								console.warn(`Invalid response for model ${ollamaModel.name}: missing required fields`)
diff --git a/src/core/webview/__tests__/webviewMessageHandler.spec.ts b/src/core/webview/__tests__/webviewMessageHandler.spec.ts
index 206c30496b..3f33dc1d3d 100644
--- a/src/core/webview/__tests__/webviewMessageHandler.spec.ts
+++ b/src/core/webview/__tests__/webviewMessageHandler.spec.ts
@@ -252,7 +252,10 @@ describe("webviewMessageHandler - requestOllamaModels", () => {
 			type: "requestOllamaModels",
 		})
 
-		expect(mockFlushModels).toHaveBeenCalledWith({ provider: "ollama", baseUrl: "http://localhost:1234" }, true)
+		expect(mockFlushModels).toHaveBeenCalledWith(
+			{ provider: "ollama", baseUrl: "http://localhost:1234", apiKey: undefined },
+			false,
+		)
 		expect(mockDiscoverOllamaModels).toHaveBeenCalledWith(
 			"http://localhost:1234",
 			undefined,
diff --git a/src/core/webview/webviewMessageHandler.ts b/src/core/webview/webviewMessageHandler.ts
index 345b7452eb..42f78e4ac1 100644
--- a/src/core/webview/webviewMessageHandler.ts
+++ b/src/core/webview/webviewMessageHandler.ts
@@ -985,12 +985,16 @@ export const webviewMessageHandler = async (
 		case "requestOllamaModels": {
 			const { apiConfiguration: ollamaApiConfig } = await provider.getState()
 			try {
-				const ollamaOptions = {
-					provider: "ollama" as const,
-					baseUrl: ollamaApiConfig.ollamaBaseUrl,
-					apiKey: ollamaApiConfig.ollamaApiKey,
-				}
-				await flushModels(ollamaOptions, true)
+				// Invalidate stale cache without refetching (discoverOllamaModelsWithSorting
+				// will perform the single authoritative API call below)
+				await flushModels(
+					{
+						provider: "ollama" as const,
+						baseUrl: ollamaApiConfig.ollamaBaseUrl,
+						apiKey: ollamaApiConfig.ollamaApiKey,
+					},
+					false,
+				)
 
 				const result = await discoverOllamaModelsWithSorting(
 					ollamaApiConfig.ollamaBaseUrl,
@@ -1004,20 +1008,19 @@ export const webviewMessageHandler = async (
 				)
 
 				// Convert modelsWithTools array to Record for compatibility
-				const modelsWithToolsRecord: Record<string, any> = {}
+				const modelsWithToolsRecord: ModelRecord = {}
 				for (const model of result.modelsWithTools) {
 					modelsWithToolsRecord[model.name] = model.modelInfo
 				}
 
-				// Always send the models message if we have any results
-				if (result.totalCount > 0) {
-					provider.postMessageToWebview({
-						type: "ollamaModels",
-						ollamaModels: modelsWithToolsRecord,
-						ollamaModelsWithTools: result.modelsWithTools,
-						modelsWithoutTools: result.modelsWithoutTools,
-					})
-				}
+				// Always send models to the webview, even when empty, so the UI
+				// can reflect that no models were found instead of silently stalling.
+				provider.postMessageToWebview({
+					type: "ollamaModels",
+					ollamaModels: modelsWithToolsRecord,
+					ollamaModelsWithTools: result.modelsWithTools,
+					modelsWithoutTools: result.modelsWithoutTools,
+				})
 			} catch (error) {
 				console.debug("Ollama models fetch failed:", error)
 			}
@@ -1065,17 +1068,16 @@ export const webviewMessageHandler = async (
 			const apiKey = message.ollamaApiKey ?? ollamaApiConfig.ollamaApiKey
 
 			try {
-				const ollamaOptions = {
-					provider: "ollama" as const,
-					baseUrl: baseUrl,
-					apiKey: apiKey,
-					ollamaModelDiscoveryTimeout: ollamaApiConfig.ollamaModelDiscoveryTimeout,
-					ollamaMaxRetries: ollamaApiConfig.ollamaMaxRetries,
-					ollamaRetryDelay: ollamaApiConfig.ollamaRetryDelay,
-					ollamaEnableLogging: ollamaApiConfig.ollamaEnableLogging,
-				}
-
-				await flushModels(ollamaOptions, true)
+				// Invalidate stale cache without refetching (discoverOllamaModelsWithSorting
+				// will perform the single authoritative API call below)
+				await flushModels(
+					{
+						provider: "ollama" as const,
+						baseUrl: baseUrl,
+						apiKey: apiKey,
+					},
+					false,
+				)
 
 				const result = await discoverOllamaModelsWithSorting(baseUrl, apiKey, {
 					modelDiscoveryTimeout: ollamaApiConfig.ollamaModelDiscoveryTimeout,
@@ -1087,7 +1089,7 @@ export const webviewMessageHandler = async (
 				const durationMs = Date.now() - startTime
 
 				// Convert modelsWithTools array to Record for compatibility
-				const modelsWithToolsRecord: Record<string, any> = {}
+				const modelsWithToolsRecord: ModelRecord = {}
 				for (const model of result.modelsWithTools) {
 					modelsWithToolsRecord[model.name] = model.modelInfo
 				}
@@ -1104,14 +1106,15 @@ export const webviewMessageHandler = async (
 					})
 				}
 
-				// Always send the models message if we have any results
+				// Always send models to the webview so the UI stays in sync
+				provider.postMessageToWebview({
+					type: "ollamaModels",
+					ollamaModels: modelsWithToolsRecord,
+					ollamaModelsWithTools: result.modelsWithTools,
+					modelsWithoutTools: result.modelsWithoutTools,
+				})
+
 				if (result.totalCount > 0) {
-					provider.postMessageToWebview({
-						type: "ollamaModels",
-						ollamaModels: modelsWithToolsRecord,
-						ollamaModelsWithTools: result.modelsWithTools,
-						modelsWithoutTools: result.modelsWithoutTools,
-					})
 					provider.postMessageToWebview({
 						type: "ollamaModelsRefreshResult",
 						success: true,