diff --git a/app/frontend/src/components/ModelManager.jsx b/app/frontend/src/components/ModelManager.jsx
index 5ef227b..c3a84ab 100644
--- a/app/frontend/src/components/ModelManager.jsx
+++ b/app/frontend/src/components/ModelManager.jsx
@@ -398,6 +398,7 @@ function ModelManager({
           backendMode: sdStatus.settings?.backendMode || sdStatus.loading?.backendMode || "",
           backendBinary: sdStatus.settings?.backendBinary || sdStatus.loading?.backendBinary || "",
           backendDevice: sdStatus.settings?.backendDevice || sdStatus.loading?.device || "",
+          llmBackendMode: llmStatus.settings?.backendMode || llmStatus.settings?.backendBinary || "",
         });
         
         if (llmStatus.ready) {
@@ -1002,6 +1003,13 @@ function ModelManager({
               <p style={{ fontSize: "0.85rem", marginTop: "2px", opacity: 0.9 }}>
                 The local C++ llama.cpp server is running. Chat requests can be sent via the Text Chat interface.
               </p>
+              {backendInfo.llmBackendMode && (
+                <div style={{ display: "flex", gap: "8px", marginTop: "8px", flexWrap: "wrap" }}>
+                  <span className="status-chip" style={{ cursor: "default", background: "rgba(255,255,255,0.15)", color: "inherit", border: "1px solid rgba(255,255,255,0.2)" }}>
+                    <span>{backendInfo.llmBackendMode}</span>
+                  </span>
+                </div>
+              )}
             </div>
           </div>
         </div>
diff --git a/app/frontend/src/components/Settings.jsx b/app/frontend/src/components/Settings.jsx
index 3b3ca4a..fd23622 100644
--- a/app/frontend/src/components/Settings.jsx
+++ b/app/frontend/src/components/Settings.jsx
@@ -1,6 +1,6 @@
-import React, { memo, useEffect } from "react";
+import React, { memo, useEffect, useState } from "react";
 import { Crop, Sliders, Cpu, Info, MessageSquare, SlidersHorizontal, Zap } from "lucide-react";
-import { stopServer, formatBytes } from "../services/api";
+import { stopServer, formatBytes, getLlmStatus } from "../services/api";
 
 const ASPECT_RATIOS = [
   { id: "1:1", label: "1:1 Square", width: 512, height: 512, sdxl_width: 1024, sdxl_height: 1024, desc: "Social posts & avatars" },
@@ -55,6 +55,26 @@ function Settings({
   const cleanupBytes = (cleanupItems || []).reduce((sum, item) => sum + Number(item.sizeBytes || 0), 0);
   const showReadinessPanel = Boolean(health && (readinessIssues.length > 0 || (cleanupItems && cleanupItems.length > 0)));
 
+  const [llmStatus, setLlmStatus] = useState({ ready: false, settings: {} });
+
+  useEffect(() => {
+    let cancelled = false;
+    const fetchLlmStatus = async () => {
+      try {
+        const status = await getLlmStatus();
+        if (!cancelled) setLlmStatus(status);
+      } catch (_) {}
+    };
+    fetchLlmStatus();
+    const interval = setInterval(fetchLlmStatus, 3000);
+    return () => {
+      cancelled = true;
+      clearInterval(interval);
+    };
+  }, []);
+
+  const supportsThinking = Boolean(llmStatus.ready && llmStatus.settings?.supportsThinking);
+
   useEffect(() => {
     if (isOpenVinoNpu && constraints.steps > 8) {
       setConstraints((prev) => ({ ...prev, steps: 8, npuSteps: 8 }));
@@ -523,7 +543,8 @@ function Settings({
                 </span>
               </div>
 
-              {/* Show Thinking Process toggle */}
+              {/* Show Thinking Process toggle - only for models that support it */}
+              {supportsThinking && (
               <div style={{ marginTop: "16px" }}>
                 <label style={{ display: "flex", alignItems: "flex-start", gap: "10px", cursor: "pointer", fontSize: "0.85rem" }}>
                   <input
@@ -546,6 +567,12 @@ function Settings({
                   </div>
                 </label>
               </div>
+              )}
+              {!supportsThinking && llmStatus.ready && (
+                <div style={{ marginTop: "16px", padding: "10px 14px", background: "rgba(251, 191, 36, 0.08)", border: "1px dashed rgb(251, 191, 36)", borderRadius: "8px", fontSize: "0.75rem", color: "var(--md-sys-color-on-surface)", lineHeight: "1.45" }}>
+                  <strong>Thinking not supported.</strong> The loaded model does not support reasoning/thinking output. DeepThink is only available for models like DeepSeek-R1, Qwen3, Gemma-4-E2B, and other reasoning-capable models.
+                </div>
+              )}
 
               {/* Temperature slider */}
               <div className="m3-slider-group" style={{ marginTop: "16px" }}>
diff --git a/app/frontend/src/components/TextChat.jsx b/app/frontend/src/components/TextChat.jsx
index 5c2c7d9..2e0ccba 100644
--- a/app/frontend/src/components/TextChat.jsx
+++ b/app/frontend/src/components/TextChat.jsx
@@ -147,6 +147,7 @@ function TextChat({
   const [attachments, setAttachments] = useState([]);
   const fileInputRef = useRef(null);
   const supportsVision = Boolean(status.ready && status.settings?.supportsVision);
+  const supportsThinking = Boolean(status.ready && status.settings?.supportsThinking);
 
   const isImage = (file) => {
     return /\.(jpe?g|png|webp)$/i.test(file.name) || file.type.startsWith("image/");
@@ -888,7 +889,7 @@ function TextChat({
                 rows={1}
               />
 
-              {status.ready && (
+              {status.ready && supportsThinking && (
                 <button
                   className={`chat-composer-deepthink-btn ${textSettings.enableThinking !== false ? "active" : ""}`}
                   onClick={() => {
diff --git a/scripts/serve.cjs b/scripts/serve.cjs
index b54e081..9ab0e3e 100644
--- a/scripts/serve.cjs
+++ b/scripts/serve.cjs
@@ -2181,6 +2181,7 @@ async function startLlm(settings = {}) {
     backendMode: backend.mode,
     backendBinary: path.basename(backend.path),
     enableThinking: settings.enableThinking !== false,
+    supportsThinking: /deepseek|qwen3|gemma-4|gemma4|think|r1|e2b|reasoning/i.test(filename),
   };
 
   let mmprojPath = null;