diff --git a/src/trpc/runtime-api.ts b/src/trpc/runtime-api.ts
index a494c86b9..04457a0a9 100644
--- a/src/trpc/runtime-api.ts
+++ b/src/trpc/runtime-api.ts
@@ -227,6 +227,20 @@ export function createRuntimeApi(deps: CreateRuntimeApiDependencies): RuntimeTrp
 								}
 							: {}),
 					});
+					const hasImages = Boolean(body.images && body.images.length > 0);
+					if (hasImages && clineLaunchConfig.modelId) {
+						const providerModels = await clineProviderService
+							.getProviderModels(clineLaunchConfig.providerId)
+							.catch(() => ({ models: [] }));
+						const selectedModel = providerModels.models.find((model) => model.id === clineLaunchConfig.modelId);
+						if (selectedModel?.supportsVision === false) {
+							return {
+								ok: false,
+								summary: null,
+								error: "The selected Cline model does not support image input. Switch to a vision-capable model or remove the images to start this task.",
+							};
+						}
+					}
 					const clineTaskSessionService = await deps.getScopedClineTaskSessionService(workspaceScope);
 					const resolvedClineTitle = resolveTaskTitle(body.taskTitle?.trim(), body.prompt);
 					const summary = await clineTaskSessionService.startTaskSession({
@@ -599,6 +613,29 @@ export function createRuntimeApi(deps: CreateRuntimeApiDependencies): RuntimeTrp
 						message: null,
 					};
 				}
+				const chatHasImages = Boolean(body.images && body.images.length > 0);
+				if (chatHasImages) {
+					try {
+						const clineLaunchConfig = await clineProviderService.resolveLaunchConfig();
+						if (clineLaunchConfig.modelId) {
+							const providerModels = await clineProviderService
+								.getProviderModels(clineLaunchConfig.providerId)
+								.catch(() => ({ models: [] }));
+							const selectedModel = providerModels.models.find(
+								(model) => model.id === clineLaunchConfig.modelId,
+							);
+							if (selectedModel?.supportsVision === false) {
+								return {
+									ok: false,
+									summary: null,
+									error: "The selected Cline model does not support image input. Switch to a vision-capable model or remove the images to send this message.",
+								};
+							}
+						}
+					} catch {
+						// Provider lookup failed — skip validation, let the SDK handle it.
+					}
+				}
 				const requestedMode = body.mode;
 				let summary = await clineTaskSessionService.sendTaskSessionInput(
 					body.taskId,
diff --git a/web-ui/src/components/detail-panels/cline-agent-chat-panel.tsx b/web-ui/src/components/detail-panels/cline-agent-chat-panel.tsx
index 1416b6e1b..9dfa8c475 100644
--- a/web-ui/src/components/detail-panels/cline-agent-chat-panel.tsx
+++ b/web-ui/src/components/detail-panels/cline-agent-chat-panel.tsx
@@ -219,7 +219,7 @@ export const ClineAgentChatPanel = React.forwardRef<ClineAgentChatPanelHandle, C
 		const panelError = composerError ?? error;
 		const attachmentWarningMessage =
 			draftImages.length > 0 && selectedModel?.supportsVision === false
-				? "The selected Cline model may not accept image input. Choose a vision-capable model to use these images."
+				? "The selected model does not support image input. Switch to a vision-capable model or remove the images to send."
 				: null;
 
 		const isPinnedToBottom = useCallback((container: HTMLDivElement): boolean => {
diff --git a/web-ui/src/components/detail-panels/cline-chat-composer.tsx b/web-ui/src/components/detail-panels/cline-chat-composer.tsx
index aa3f09a68..19b386dec 100644
--- a/web-ui/src/components/detail-panels/cline-chat-composer.tsx
+++ b/web-ui/src/components/detail-panels/cline-chat-composer.tsx
@@ -113,7 +113,7 @@ export function ClineChatComposer({
 	const [slashSuggestions, setSlashSuggestions] = useState<ClineComposerCompletionSuggestion[]>([]);
 	const [isMentionSearchLoading, setIsMentionSearchLoading] = useState(false);
 	const [isSlashSearchLoading, setIsSlashSearchLoading] = useState(false);
-	const canSubmit = canSend && !isModelSaving && (draft.trim().length > 0 || images.length > 0);
+	const canSubmit = canSend && !isModelSaving && !attachmentWarningMessage && (draft.trim().length > 0 || images.length > 0);
 
 	const activeToken = useMemo(() => detectActiveClineComposerToken(draft, cursorIndex), [cursorIndex, draft]);
 	const completionSuggestions = useMemo(() => {
diff --git a/web-ui/src/components/task-create-dialog.tsx b/web-ui/src/components/task-create-dialog.tsx
index 6020693b5..9e7ad7acd 100644
--- a/web-ui/src/components/task-create-dialog.tsx
+++ b/web-ui/src/components/task-create-dialog.tsx
@@ -3,6 +3,7 @@ import * as DropdownMenu from "@radix-ui/react-dropdown-menu";
 import * as RadixSwitch from "@radix-ui/react-switch";
 
 import {
+	AlertTriangle,
 	ArrowBigUp,
 	ArrowLeft,
 	Check,
@@ -202,6 +203,18 @@ export function TaskCreateDialog({
 		onCreateStartAndOpen || primaryStartAction === "start" ? primaryStartAction : DEFAULT_PRIMARY_START_ACTION;
 	const secondaryStartAction = effectivePrimaryStartAction === "start" ? "start_and_open" : "start";
 
+	const effectiveAgentIdForImages = (agentId || defaultAgentId) ?? null;
+	const effectiveModelIdForImages = clineSettings?.modelId ?? effectiveDefaultModelId ?? "";
+	const selectedClineModel = useMemo(
+		() =>
+			effectiveModelIdForImages
+				? (providerModels.find((model) => model.id === effectiveModelIdForImages) ?? null)
+				: null,
+		[effectiveModelIdForImages, providerModels],
+	);
+	const imagesBlockedByModel =
+		images.length > 0 && effectiveAgentIdForImages === "cline" && selectedClineModel?.supportsVision === false;
+
 	// Reset state when dialog closes
 	useEffect(() => {
 		if (!open) {
@@ -598,6 +611,15 @@ export function TaskCreateDialog({
 				</div>
 			</DialogBody>
 			<DialogFooter>
+				{imagesBlockedByModel ? (
+					<div className="mb-2 flex items-start gap-1.5 text-xs text-status-orange">
+						<AlertTriangle size={14} className="mt-0.5 shrink-0" />
+						<p className="m-0 min-w-0">
+							The selected Cline model does not support image input. Switch to a vision-capable model in Override
+							Agent Settings or remove the images to continue.
+						</p>
+					</div>
+				) : null}
 				<label
 					htmlFor={createMoreId}
 					className="mr-auto flex items-center gap-2 text-[12px] text-text-primary cursor-pointer select-none"
@@ -614,7 +636,11 @@ export function TaskCreateDialog({
 				</label>
 				{mode === "single" ? (
 					<>
-						<Button size="sm" onClick={handleCreateSingle} disabled={!prompt.trim() || !branchRef}>
+						<Button
+							size="sm"
+							onClick={handleCreateSingle}
+							disabled={!prompt.trim() || !branchRef || imagesBlockedByModel}
+						>
 							<span className="inline-flex items-center">
 								Create
 								<ButtonShortcut />
@@ -627,7 +653,7 @@ export function TaskCreateDialog({
 										variant="primary"
 										size="sm"
 										onClick={() => handleRunSingleStartAction(primaryStartAction)}
-										disabled={!prompt.trim() || !branchRef}
+										disabled={!prompt.trim() || !branchRef || imagesBlockedByModel}
 										className={onCreateStartAndOpen ? "rounded-r-none" : undefined}
 									>
 										<span className="inline-flex items-center">
@@ -640,7 +666,7 @@ export function TaskCreateDialog({
 											<Button
 												variant="primary"
 												size="sm"
-												disabled={!prompt.trim() || !branchRef}
+												disabled={!prompt.trim() || !branchRef || imagesBlockedByModel}
 												className="rounded-l-none border-l border-white/20 px-1"
 												aria-label="More start options"
 											>