Serve Humanlike Gemma: model gemma-4-31b-humanlike-chat, thinking on/off, title Humanlike Chat

README.md

@@ -1,27 +1,23 @@
 ---
-title: Qwen3.8 27B Humanlike Chat
+title: Humanlike Chat
 colorFrom: indigo
 colorTo: pink
 sdk: static
 app_file: index.html
 pinned: false
 fullWidth: true
-short_description: Chat with the hosted Qwen3.8 Humanlike endpoint.
-models:
-  - LessThanThreeAI/Qwen3.8-27B-Humanlike-Chat-GGUF
+short_description: Chat with the hosted Humanlike model.
 ---
 
-# Qwen3.8 27B Humanlike Chat
+# Humanlike Chat
 
 Static chat client for the hosted OpenAI-compatible endpoint at `https://api.lessthanthreeai.com/v1`.
 
-- Model: `qwen3.8-27b-humanlike-chat`
+- Model: `gemma-4-31b-humanlike-chat`
 - No API key required
 - Optional system prompt for a persona, setting, or relationship, kept in the browser between reloads
-- Reasoning level selector: off (default), low, medium, or xhigh
+- Thinking: off (default) or on
 - One conversation per page load; refreshing starts a new session
 - Requests send only `model`, `messages`, `stream`, `max_tokens`, and `chat_template_kwargs`, so the endpoint defaults match the deployment gate
 
-The service runs on scale-to-zero capacity. The first message after an idle period can take about two to three minutes while a worker starts, and the request stays connected while that happens. Later messages are fast until the worker shuts down after five idle minutes.
-
-See the [model card](https://huggingface.co/LessThanThreeAI/Qwen3.8-27B-Humanlike-Chat-GGUF) for the artifacts and the [serving configuration](https://huggingface.co/LessThanThreeAI/Qwen3.8-27B-Humanlike-Chat-GGUF/blob/main/SERVING.md) for the engine setup.
+The service runs on scale-to-zero capacity. The first message after an idle period can take a few minutes while a worker starts, and the request stays connected while that happens. Later messages are fast until the worker shuts down after five idle minutes.

index.html

@@ -3,7 +3,7 @@
 <head>
   <meta charset="utf-8">
   <meta name="viewport" content="width=device-width, initial-scale=1">
-  <title>Qwen3.8-27B Humanlike Chat</title>
+  <title>Humanlike Chat</title>
   <style>
     * { box-sizing: border-box; }
     body {
@@ -137,10 +137,9 @@
 <body>
   <main>
     <header>
-      <h1>Qwen3.8-27B Humanlike Chat</h1>
+      <h1>Humanlike Chat</h1>
       <p>
         Live demo of the hosted endpoint <code>https://api.lessthanthreeai.com/v1</code>.
-        <a href="https://huggingface.co/LessThanThreeAI/Qwen3.8-27B-Humanlike-Chat-GGUF">Model card</a>
       </p>
       <details>
         <summary>Settings</summary>
@@ -150,12 +149,10 @@
             placeholder="Optional. Sets identity, setting, or relationship for the conversation."></textarea>
         </div>
         <div class="field">
-          <label for="reasoning">Reasoning</label>
+          <label for="reasoning">Thinking</label>
           <select id="reasoning">
             <option value="off">Off (fastest, the default)</option>
-            <option value="low">Low</option>
-            <option value="medium">Medium</option>
-            <option value="xhigh">Extra high</option>
+            <option value="on">On</option>
           </select>
         </div>
       </details>
@@ -173,9 +170,9 @@
 
   <script>
     const API = "https://api.lessthanthreeai.com/v1";
-    const MODEL = "qwen3.8-27b-humanlike-chat";
-    const INTRO = "Say something. The first reply can take a couple of minutes while the inference worker starts.";
-    const MAX_TOKENS = { off: 512, low: 4096, medium: 4096, xhigh: 8192 };
+    const MODEL = "gemma-4-31b-humanlike-chat";
+    const INTRO = "Say something. The first reply can take a few minutes while the inference worker starts.";
+    const MAX_TOKENS = { off: 512, on: 4096 };
 
     const chat = document.querySelector("#chat");
     const form = document.querySelector("#form");
@@ -189,9 +186,9 @@
     const messages = [];
 
     system.value = localStorage.getItem("systemPrompt") || "";
-    reasoning.value = localStorage.getItem("reasoningEffort") || "off";
+    reasoning.value = localStorage.getItem("thinking") === "on" ? "on" : "off";
     system.addEventListener("input", () => localStorage.setItem("systemPrompt", system.value));
-    reasoning.addEventListener("change", () => localStorage.setItem("reasoningEffort", reasoning.value));
+    reasoning.addEventListener("change", () => localStorage.setItem("thinking", reasoning.value));
 
     function addMessage(role, text) {
       const element = document.createElement("div");
@@ -259,7 +256,7 @@
       prompt.value = "";
       send.disabled = true;
       status.classList.add("busy");
-      status.textContent = "Starting the model. A cold start takes about two to three minutes.";
+      status.textContent = "Starting the model. A cold start can take a few minutes.";
 
       const target = addAssistantMessage();
       const effort = reasoning.value;
@@ -269,9 +266,7 @@
         ? [{ role: "system", content: systemPrompt }, ...messages]
         : messages;
 
-      const chatTemplateKwargs = effort === "off"
-        ? { enable_thinking: false, preserve_thinking: false }
-        : { enable_thinking: true, reasoning_effort: effort, preserve_thinking: false };
+      const chatTemplateKwargs = { enable_thinking: effort === "on", preserve_thinking: false };
 
       try {
         const response = await fetch(`${API}/chat/completions`, {