Magistral-2509-36B-Text-Only

<style> body { font-family: 'Quicksand', sans-serif; /* Replaced purple gradient with a warm, fiery one */ background: linear-gradient(135deg, #4a1e00 0%, #1c0a00 100%); /* Changed text color to a warmer, parchment-like off-white */ color: #F5EFE6; margin: 0; padding: 0; font-size: 16px; } h1, h2, h3, h4, summary { font-family: 'Cinzel', serif; } .container { margin: 20px auto; max-width: 900px; background-color: rgba(28, 22, 18, 0.95); padding: 30px; border-radius: 12px; box-shadow: 0 4px 20px rgba(255, 140, 0, 0.15); border: 1px solid rgba(255, 140, 0, 0.2); outline: 1px solid rgba(255, 140, 0, 0.5); outline-offset: -1px; position: relative; } .container::before { content: ''; position: absolute; top: -1px; left: -1px; right: -1px; bottom: -1px; border: 1px solid rgba(255, 165, 0, 0.98); border-radius: 12px; pointer-events: none; animation: borderGlow 2.5s ease-in-out infinite; } @keyframes borderGlow { 0% { box-shadow: 0 0 5px rgba(255, 165, 0, 0.98); } 50% { box-shadow: 0 0 12px rgba(255, 165, 0, 0.98); } 100% { box-shadow: 0 0 5px rgba(255, 165, 0, 0.98); } } .header h1 { font-size: 32px; color: #FFA500; margin: 0 0 20px 0; text-align: center; text-shadow: 0 0 12px rgba(255, 100, 0, 0.6); } a { color: #FFD700; text-decoration: none; transition: color 0.3s ease; } a:hover { color: #FFDAB9; } .button { display: inline-block; background-color: #E55B00; color: #FFFFFF; padding: 12px 24px; border-radius: 5px; cursor: pointer; text-decoration: none; font-family: 'Cinzel', serif; font-weight: 600; transition: all 0.3s ease; border: 1px solid transparent; } .button:hover { background-color: #FF8C00; box-shadow: 0 0 15px rgba(255, 140, 0, 0.5); transform: translateY(-2px); } pre { background-color: rgba(45, 35, 25, 0.95); padding: 15px; border-radius: 5px; overflow-x: auto; border: 1px solid rgba(255, 140, 0, 0.2); outline: 1px solid rgba(255, 140, 0, 0.5); outline-offset: -1px; } code { font-family: 'Courier New', monospace; /* Code text uses the new base text color */ color: #F5EFE6; } /* Section Container */ .section-container { margin: 40px 0; } h2 { font-size: 26px; /* Section headers are orange */ color: #FFA500; text-shadow: 0 0 10px rgba(255, 140, 0, 0.5); border-bottom: 1px solid rgba(255, 140, 0, 0.2); padding-bottom: 10px; margin-bottom: 20px; } .info-card { /* Card background is a warm dark brown */ background: rgba(45, 35, 25, 0.95); border: 1px solid rgba(255, 140, 0, 0.2); border-radius: 8px; overflow: hidden; margin-bottom: 25px; } .info-header { /* Header background has an orange tint */ background: rgba(255, 140, 0, 0.1); padding: 20px; border-bottom: 1px solid rgba(255, 140, 0, 0.2); } .info-header h3 { /* Card titles are orange */ color: #FFA500; margin: 0 0 10px 0; font-size: 22px; text-shadow: 0 0 5px rgba(255, 140, 0, 0.3); } .card-content { padding: 20px; line-height: 1.7; } .card-content p, .card-content li { margin-bottom: 1em; } .card-content p:last-child, .card-content li:last-child { margin-bottom: 0; } .card-content ul { list-style: none; padding-left: 20px; } .card-content li::before { content: '✦'; /* Bullet points are gold */ color: #FFD700; font-weight: bold; display: inline-block; width: 1em; margin-left: -1.2em; font-size: 1.2em; line-height: 1; } .card-content strong { /* Strong text is gold */ color: #FFD700; font-weight: 600; } /* Configuration */ .config-container { background: rgba(45, 35, 25, 0.95); border: 1px solid rgba(255, 140, 0, 0.2); border-radius: 8px; overflow: hidden; } .config-header { background: rgba(255, 140, 0, 0.1); padding: 15px 20px; border-bottom: 1px solid rgba(255, 140, 0, 0.2); } .config-header h3 { margin: 0; color: #FFA500; font-size: 22px; } .config-content { padding: 20px; display: grid; grid-template-columns: repeat(auto-fit, minmax(200px, 1fr)); gap: 20px; } .config-item { display: flex; flex-direction: column; gap: 5px; } .config-label { /* Config labels are gold */ color: #FFD700; font-size: 14px; font-weight: 500; font-family: 'Quicksand', sans-serif; } .config-value { /* Config values use the new base text color */ color: #F5EFE6; font-family: 'Courier New', monospace; font-size: 18px; font-weight: bold; } .support-section { text-align: center; margin-top: 40px; background: rgba(45, 35, 25, 0.95); border: 1px solid rgba(255, 140, 0, 0.2); border-radius: 8px; padding: 20px; } .support-section p { margin-bottom: 15px; font-size: 1.1em; margin-top: 0; } summary { cursor: pointer; list-style: none; /* Remove default arrow */ outline: none; display: flex; align-items: flex-start; /* Align items to the top to respect h2's vertical space */ } summary::-webkit-details-marker { display: none; } summary::before { content: '▶'; font-size: 1.2em; color: #FFA500; /* Match h2 color */ margin-right: 15px; padding-top: 5px; /* Adjust vertical alignment with h2 text */ transition: transform 0.2s ease; flex-shrink: 0; /* Prevent the arrow from shrinking */ } details[open] > summary::before { transform: rotate(90deg); } summary > h2 { flex-grow: 1; } .format-list { margin: 1.5em 0; } .format-list dt { color: #FFA500; /* Your primary orange header color */ font-family: 'Cinzel', serif; /* Your header font for consistency */ font-weight: 600; /* Ensure emphasis */ font-size: 1.3em; /* Make it more prominent */ text-shadow: 0 0 8px rgba(255, 140, 0, 0.5); /* Add the signature glow */ margin-bottom: 0.6em; /* A little more space underneath */ } .format-list dd { /* The example code block */ margin-left: 0; margin-bottom: 1.5em; background-color: rgba(28, 22, 18, 0.95); padding: 12px 15px; border-radius: 6px; border: 1px solid rgba(255, 140, 0, 0.2); border-left: 4px solid #FFA500; } .format-list dd:last-of-type { margin-bottom: 0; } table { width: 100%; border-collapse: collapse; margin: 25px 0; font-size: 1em; font-family: 'Quicksand', sans-serif; min-width: 400px; border-radius: 5px 5px 0 0; overflow: hidden; box-shadow: 0 0 20px rgba(0, 0, 0, 0.15); } table thead tr { background-color: #FFA500; color: #1c0a00; text-align: left; font-weight: bold; font-family: 'Cinzel', serif; } table th, table td { padding: 12px 15px; border-bottom: 1px solid rgba(255, 140, 0, 0.2); } table tbody tr { background-color: rgba(45, 35, 25, 0.95); } table tbody tr:last-of-type { border-bottom: 2px solid #FFA500; } table tbody tr.active-row { font-weight: bold; color: #FFA500; } </style> <div class="container"> <link href="https://fonts.googleapis.com/css2?family=Cinzel:wght@400;500;600&family=Quicksand:wght@400;500&display=swap" rel="stylesheet"> <div class="header"> <h1>Magistral-Small-2509-36B-Text-Only</h1> </div> <div class="info"> <div class="support-section"> <p><strong>Send me your support to help me feed the data beast! also taking comissions for universe specific models</strong></p> <a href="https://ko-fi.com/som1tokmynam" target="_blank" class="button"> Support on Ko-fi </a> </div> <div class="section-container"> <details open> <summary><h2>Model Description</h2></summary> <div class="info-card"> <div class="card-content"> <p>This model is a <strong>36B parameter upscale</strong> of the popular <code>Darkhn/Magistral-Small-2509-Text-Only</code> model. The goal of this upscale is to further enhance the powerful reasoning and knowledge-based capabilities of the original text-only model by increasing its parameter count.</p> <p>The original model was a text-only version of Mistral AI's <strong>Magistral Small 1.2</strong> (built on Mistral Small 3.2, 2506), created by removing the vision encoder. This 36B version builds directly upon that foundation, inheriting its deep capacity for logical deduction and problem-solving, and aims to provide even deeper results while remaining exclusively focused on text-based tasks.</p> </div> </div> </details> </div> <div class="section-container"> <details open> <summary><h2>Important: Reasoning Format & Backend Setup</h2></summary> <div class="info-card"> <div class="card-content"> <p>This model uses a special reasoning format. There are two methods to enable it: the <strong>official format</strong> designed by MistralAI, and a <strong>legacy format</strong> that works due to the base model's pre-training. The correct method depends on your backend software (e.g., llama.cpp, Kobold.cpp).</p> <hr> <h3>Official Format: <code>[THINK]</code> (Recommended for llama.cpp)</h3> <p>This is the official instruction format from MistralAI and is the recommended method. It is confirmed to work with backends like <strong>llama.cpp</strong> (with specific flags) and <strong>mistral-common</strong>.</p> <ul> <li><strong>Llama.cpp Prerequisite:</strong> Launch llama.cpp with the <code>--special</code> and <code>--jinja</code> arguments enabled.</li> <li><strong>Instruction Format:</strong> The model uses <code>[THINK]</code> and <code>[/THINK]</code> tags.</li> <li><strong>Activation (2 steps):</strong> <ol> <li>Set your prefill sequence (in your frontend like SillyTavern) to start with <code>[THINK]</code>.</li> <li>You <strong>must</strong> also include the keyword <code>/think</code> anywhere in your system prompt to activate the reasoning module.</li> </ol> </li> </ul> <h4>Recommended System Prompt for Official Format</h4> <p>Add the following to your system prompt to guide the model's output structure:</p> <pre><code>First draft your thinking process (inner monologue) until you arrive at a response. You must use the /think keyword. Format your response using Markdown, and use LaTeX for any mathematical equations. Write both your thoughts and the response in the same language as the input. Your thinking process must follow the template below:[THINK]Your thoughts or/and draft, like working through an exercise on scratch paper. Be as casual and as long as you want until you are confident to generate the response. Use the same language as the input.[/THINK]Here, provide a self-contained response.</code></pre> <h4>SillyTavern Quick Setup</h4> <p>For a complete SillyTavern configuration, you can download and import this JSON file:</p> <a href="https://huggingface.co/Darkhn/Sampler_settings_and_system_prompt/blob/main/Magistral_SillyTavern_settings.json" target="_blank" class="button"> Download SillyTavern JSON <span class="link-arrow">→</span> </a> <hr> <h3>Legacy Format: <code>&lt;think&gt;</code> (For Kobold.cpp & TabbyAPI)</h3> <p>This format is not official but is highly effective with backends like <strong>Kobold.cpp</strong> and <strong>TabbyAPI</strong>. It works because the model's predecessor was trained on these angle-bracket tags, and the model inherits this behavior.</p> <ul> <li><strong>Instruction Format:</strong> Wrap the model's reasoning in <code>&lt;think&gt;</code> and <code>&lt;/think&gt;</code> tags.</li> <li><strong>Activation:</strong> In your frontend, set your prefill sequence to start with <code>&lt;think&gt;</code>.</li> </ul> <p> <a href="https://github.com/LostRuins/koboldcpp/issues/1745" target="_blank" class="button"> See the GitHub Issue for technical details <span class="link-arrow">→</span> </a> </p> </div> </div> </details> </div> <div class="section-container"> <details open> <summary><h2>Key Features</h2></summary> <div class="info-card"> <div class="card-content"> <ul> <li><strong>Reasoning:</strong> Capable of producing long, coherent chains of thought to break down complex problems before providing an answer.</li> <li><strong>Multilingual:</strong> Supports dozens of languages, including English, French, German, Spanish, Italian, Japanese, Korean, Chinese, Arabic, and many more.</li> <li><strong>Apache 2.0 License:</strong> Features a permissive, open license allowing for both commercial and non-commercial use.</li> <li><strong>Context Window:</strong> A 128k context window. Performance might degrade past <strong>40k</strong>, but the model should still provide good results.</li> </ul> </div> </div> </details> </div> <div class="section-container"> <details open> <summary> <h2>Usage Guide</h2> </summary> <div class="info-card"> <div class="info-header"> <h3>Recommended Sampler Settings</h3> </div> <div class="config-container"> <div class="config-content"> <div class="config-item"> <span class="config-label">Temperature</span> <span class="config-value">0.7</span> </div> <div class="config-item"> <span class="config-label">Top P</span> <span class="config-value">0.95</span> </div> <div class="config-item"> <span class="config-label">Max Tokens</span> <span class="config-value">131072</span> </div> </div> </div> </div> <div class="info-card"> <div class="card-content"> <p>For best results, you must use the recommended system prompt and response format. The model uses special <code>[THINK]</code> and <code>[/THINK]</code> tokens to encapsulate its reasoning process before delivering the final answer.</p> <dl class="format-list"> <dt>System Prompt</dt> <dd><code>First draft your thinking process (inner monologue) until you arrive at a response. Format your response using Markdown, and use LaTeX for any mathematical equations. Write both your thoughts and the response in the same language as the input.<br><br>Your thinking process must follow the template below:[THINK]Your thoughts or/and draft, like working through an exercise on scratch paper. Be as casual and as long as you want until you are confident to generate the response. Use the same language as the input.[/THINK]Here, provide a self-contained response.</code></dd> </dl> <p><strong>Important:</strong> The <code>[THINK]</code> and <code>[/THINK]</code> tags are special tokens and must be encoded as such. Please ensure you are using a recent version of a library that supports the Magistral chat template, such as <code>mistral-common</code>.</p> </div> </div> </details> </div> <div class="section-container"> <details open> <summary><h2>Benchmark Results</h2></summary> <div class="info-card"> <div class="card-content"> <p>The following benchmarks are from the official release of the original multimodal <strong>Magistral Small 1.2</strong>. While this 36B upscale is derived from that lineage, its performance will differ due to the significant increase in parameter count. These scores are provided for historical context of the base architecture.</p> <table> <thead> <tr> <th>Model</th> <th>AIME24 pass@1</th> <th>AIME25 pass@1</th> <th>GPQA Diamond</th> <th>Livecodebench (v5)</th> </tr> </thead> <tbody> <tr> <td>Magistral Medium 1.2</td> <td><b>91.82%</b></td> <td><b>83.48%</b></td> <td><b>76.26%</b></td> <td><b>75.00%</b></td> </tr> <tr> <td>Magistral Small 1.2</td> <td><b>86.14%</b></td> <td><b>77.34%</b></td> <td><b>70.07%</b></td> <td><b>70.88%</b></td> </tr> <tr> <td>Magistral Small 1.1</td> <td>70.52%</td> <td>62.03%</td> <td>65.78%</td> <td>59.17%</td> </tr> </tbody> </table> </div> </div> </details> </div> <div class="section-container"> <details> <summary><h2>Intended Use & Limitations</h2></summary> <div class="info-card"> <div class="card-content"> <ul> <li><strong>Intended Use:</strong> This model is designed for text-based tasks that require strong reasoning, instruction following, and multilingual chat capabilities, without the computational overhead of multimodal features.</li> <li><strong>Limitations & Quirks:</strong> <ul> <li>This is a <strong>text-only model</strong> and cannot process image or other non-text inputs.</li> <li>Performance on tasks outside of its core training domain (e.g., highly specialized coding, non-chat formats) is not guaranteed.</li> <li>The model may "hallucinate" or generate plausible but incorrect information. Always verify critical facts.</li> <li><strong>Safety:</strong> This model has not undergone additional safety alignment beyond what was included in its base Magistral model. Standard responsible AI practices should be followed.</li> </ul> </li> </ul> </div> </div> </details> </div> <div class="section-container"> <details> <summary><h2>Acknowledgements</h2></summary> <div class="info-card"> <div class="card-content"> <ul> <li>Credit to Mistral AI for the powerful Magistral architecture and for releasing their work openly.</li> </ul> </div> </div> </details> </div> </div> </div>