diff --git a/public/app/constants.js b/public/app/constants.js
index 478d154..0b63eea 100644
--- a/public/app/constants.js
+++ b/public/app/constants.js
@@ -14,8 +14,3 @@ export const themes = {
     "beta-ai": {name: "Beta AI", short: "BA"},
     "confusity": {name: "Confusity", short: "CF"}
 };
-
-export const defaultModelProfile = {
-    badge: "General",
-    summary: "General-purpose chat model. Start here if you are unsure which model to use."
-};
diff --git a/public/app/modelProfiles.js b/public/app/modelProfiles.js
new file mode 100644
index 0000000..360d3e6
--- /dev/null
+++ b/public/app/modelProfiles.js
@@ -0,0 +1,89 @@
+export const defaultModelProfile = {
+    badge: "General",
+    summary: "Best for general questions, drafting, and everyday conversation. Use a specialized model when you need stronger coding, reasoning, or roleplay behavior."
+};
+
+const modelProfileRules = [
+    {
+        matches: ["catgirl"],
+        badge: "Roleplay",
+        summary: "Best for playful catgirl character roleplay and expressive, in-character conversation. Not intended for neutral factual answers or technical work."
+    },
+    {
+        matches: ["femboy"],
+        badge: "Roleplay",
+        summary: "Best for personality-led roleplay, casual conversation, and maintaining a stylized character voice. Not intended for precise research or technical tasks."
+    },
+    {
+        matches: ["buenzli", "bünzli"],
+        badge: "Roleplay",
+        summary: "Best for Swiss-flavored character roleplay, local humor, and casual conversation. Choose a general model for neutral writing or factual questions."
+    },
+    {
+        matches: ["deepseek-r1", "deepseek-r1-distill"],
+        badge: "Reasoning",
+        summary: "Best for problems that benefit from deliberate reasoning, such as logic, mathematics, planning, and difficult technical questions. It can be slower and more verbose for simple chat."
+    },
+    {
+        matches: ["codellama", "code-llama", "deepseek-coder", "qwen2.5-coder", "qwen3-coder", "starcoder", "codestral", "coder"],
+        badge: "Coding",
+        summary: "Best for writing, explaining, reviewing, and debugging code. Choose a general model for creative conversation or broad non-technical questions."
+    },
+    {
+        matches: ["dolphin"],
+        badge: "Creative",
+        summary: "Best for open-ended conversation, brainstorming, creative writing, and roleplay. Its answers can be less restrained and less dependable for factual or safety-critical topics."
+    },
+    {
+        matches: ["mixtral"],
+        badge: "Versatile",
+        summary: "Best for detailed writing, summarization, multilingual tasks, and general problem-solving. It is a capable all-rounder but may respond more slowly than smaller models."
+    },
+    {
+        matches: ["mistral"],
+        badge: "Balanced",
+        summary: "Best for everyday questions, concise writing, summaries, and light coding. A strong default when you want a balance of speed and answer quality."
+    },
+    {
+        matches: ["gemma"],
+        badge: "Efficient",
+        summary: "Best for quick answers, rewriting, summarization, and lightweight general chat. Larger or reasoning-focused models are better for complex multi-step problems."
+    },
+    {
+        matches: ["qwen"],
+        badge: "Versatile",
+        summary: "Best for multilingual conversation, structured writing, instruction following, and general technical tasks. Coder variants are the better choice for code-heavy work."
+    },
+    {
+        matches: ["llama"],
+        badge: "General",
+        summary: "Best for general conversation, writing, summarization, and broad knowledge tasks. Specialized coding or reasoning models may perform better in those areas."
+    },
+    {
+        matches: ["phi"],
+        badge: "Compact",
+        summary: "Best for fast, focused questions, short summaries, and lightweight assistance. Its compact size makes it less suitable for demanding reasoning or long, nuanced responses."
+    }
+];
+
+function describeBuild(model) {
+    const parameterSize = String(model?.details?.parameter_size || "").trim();
+    const quantization = String(model?.details?.quantization_level || "").trim();
+    const details = [parameterSize, quantization].filter(Boolean);
+    return details.length ? ` Available build: ${details.join(", ")}.` : "";
+}
+
+export function getModelProfile(model = {}) {
+    const modelId = String(model.model || model.name || "");
+    const modelName = String(model.name || model.model || "");
+    const family = String(model?.details?.family || "");
+    const families = Array.isArray(model?.details?.families) ? model.details.families.join(" ") : "";
+    const haystack = `${modelId} ${modelName} ${family} ${families}`.toLowerCase();
+    const match = modelProfileRules.find((profile) => profile.matches.some((token) => haystack.includes(token)));
+    const profile = match || defaultModelProfile;
+
+    return {
+        badge: profile.badge,
+        summary: `${profile.summary}${describeBuild(model)}`
+    };
+}
diff --git a/public/index.html b/public/index.html
index 385abc1..8d3532b 100644
--- a/public/index.html
+++ b/public/index.html
@@ -424,6 +424,6 @@
 
 <script src="https://cdn.jsdelivr.net/npm/marked/marked.min.js" defer></script>
 <script src="https://cdn.jsdelivr.net/npm/dompurify@3.0.6/dist/purify.min.js" defer></script>
-<script type="module" src="script.js?v=20260908-model-controls"></script>
+<script type="module" src="script.js?v=20260908-model-descriptions"></script>
 </body>
 </html>
diff --git a/public/script.js b/public/script.js
index f9cfbc4..20e00d8 100644
--- a/public/script.js
+++ b/public/script.js
@@ -1,6 +1,7 @@
 import {del, get, post, put, stream} from "./app/api.js";
 import {closeChatDrawer, openChatDrawer} from "./app/chatDrawer.js";
-import {defaultModelProfile, onboardingPrompts, requestedChatId} from "./app/constants.js";
+import {onboardingPrompts, requestedChatId} from "./app/constants.js";
+import {getModelProfile} from "./app/modelProfiles.js";
 import {applyTheme as applySharedTheme, applyThemeMode as applySharedThemeMode, readStoredAppearance} from "./app/themeController.js";
 import {
     activeChatTitle,
@@ -129,6 +130,7 @@ let onboardingStep = 1, onboardingIntent = "ask";
 let chatLoadingDepth = 0;
 let chatActivityDepth = 0;
 let messageRetryState = new Map();
+let availableModelsById = new Map();
 
 
 function setAuthMessage(message, state = "") { authMsg.textContent = message; authMsg.className = state ? `status ${state}` : "status"; }
@@ -918,41 +920,6 @@ function renderModelSection() {
     return modelSelect?.selectedOptions[0]?.textContent || "";
 }
 
-function getModelProfile(modelId, modelName) {
-    const haystack = `${modelId || ""} ${modelName || ""}`.toLowerCase();
-    if (haystack.includes("codellama") || haystack.includes("code")) {
-        return {
-            badge: "Coding",
-            summary: "Best for code generation, debugging, and technical explanations. Usually stronger on programming tasks than on creative chat."
-        };
-    }
-    if (haystack.includes("gemma")) {
-        return {
-            badge: "Fast",
-            summary: "A lighter general chat model. Good for quick answers and lower-latency replies, with less depth than larger models."
-        };
-    }
-    if (haystack.includes("dolphin")) {
-        return {
-            badge: "Chatty",
-            summary: "Instruction-following conversational model. Good for open-ended chat, brainstorming, and longer natural responses."
-        };
-    }
-    if (haystack.includes("mistral")) {
-        return {
-            badge: "Balanced",
-            summary: "Strong default for everyday use. Usually a good balance between speed, clarity, and response quality."
-        };
-    }
-    if (haystack.includes("catgirl") || haystack.includes("femboy") || haystack.includes("buenzli")) {
-        return {
-            badge: "Persona",
-            summary: "Specialized character-style model. Best for roleplay or stylized voice, and less reliable for factual or neutral answers."
-        };
-    }
-    return defaultModelProfile;
-}
-
 function updateModelHelp() {
     if (!modelSelect || !modelHelpTitle || !modelHelpBadge || !modelHelpSummary) return;
     const selectedOption = modelSelect.selectedOptions[0];
@@ -972,7 +939,7 @@ function updateModelHelp() {
     }
     const modelId = selectedOption.value;
     const modelName = selectedOption.textContent || modelId;
-    const profile = getModelProfile(modelId, modelName);
+    const profile = getModelProfile(availableModelsById.get(modelId) || {model: modelId, name: modelName});
     if (modelBadgeName) modelBadgeName.textContent = modelName;
     if (headerModelName) headerModelName.textContent = modelName;
     modelHelpTitle.textContent = modelName;
@@ -1810,22 +1777,26 @@ async function displayModels() {
     const res = await get("/models");
     const models = Array.isArray(res.models) ? res.models : null;
     if (res.error || !models) {
+        availableModelsById = new Map();
         modelSelect.innerHTML = "<option>Error loading models</option>";
         renderModelSection();
         return setNotice("Models could not be loaded.", "error");
     }
     modelSelect.innerHTML = "";
+    availableModelsById = new Map(models.map((model) => [String(model.model || model.name || ""), model]));
     const preferredModelId = "mistral:latest";
-    const preferredIndex = models.findIndex((model) => model.model === preferredModelId);
+    const preferredIndex = models.findIndex((model) => (model.model || model.name) === preferredModelId);
     models.forEach((model, index) => {
+        const modelId = String(model.model || model.name || "");
         const option = document.createElement("option");
-        option.value = model.model;
-        option.textContent = model.name;
+        option.value = modelId;
+        option.textContent = model.name || modelId;
+        option.title = getModelProfile(model).summary;
         option.selected = preferredIndex >= 0 ? index === preferredIndex : index === 0;
         modelSelect.appendChild(option);
     });
     const activeChat = getChatById(activeChatId);
-    if (activeChat?.preferred_model && models.some((model) => model.model === activeChat.preferred_model)) {
+    if (activeChat?.preferred_model && models.some((model) => (model.model || model.name) === activeChat.preferred_model)) {
         modelSelect.value = activeChat.preferred_model;
     }
     renderModelSection();
diff --git a/scripts/lint.mjs b/scripts/lint.mjs
index 15b09a9..5046922 100644
--- a/scripts/lint.mjs
+++ b/scripts/lint.mjs
@@ -24,7 +24,8 @@ const files = [
     "public/app/api.js",
     "public/app/chatDrawer.js",
     "public/app/constants.js",
-    "public/app/dom.js"
+    "public/app/dom.js",
+    "public/app/modelProfiles.js"
 ];
 
 for (const file of files) {