xeyonai/mistral-helcyon-4o-12b-v1.0-gguf overview

{
  "metadata": {},
  "card_data": {
    "license": "apache-2.0",
    "language": [
      "en"
    ],
    "base_model": [
      "mistralai/Mistral-Nemo-Instruct-2407"
    ],
    "tags": [
      "companion",
      "assistant",
      "conversational",
      "roleplay",
      "adventure",
      "writing",
      "long-context"
    ],
    "frontmatter": {
      "license": "apache-2.0",
      "language": [
        "en"
      ],
      "base_model": [
        "mistralai/Mistral-Nemo-Instruct-2407"
      ],
      "tags": [
        "companion",
        "assistant",
        "conversational",
        "roleplay",
        "adventure",
        "writing",
        "long-context"
      ]
    },
    "hero_image_url": "https://cdn-uploads.huggingface.co/production/uploads/6951bbce66023f270e224d0d/ZkgdN5ACln02KDf-xiwzC.png",
    "summary": "**Model Name:** helcyon-4o-12b-GGUF **Version:** 4.0 **Owner:** HardWire **Base:** Mistral Nemo 12B (full weight trained + LoRA merges) **Quantized GGUFs:** Q3_K_M, Q4_K_M, Q5_K_M, Q6_K, Q8_0 **Tags:** local-llm, conversational, companion, emotional-intelligence, long-context, roleplay, creative-writing ---",
    "quick_links": [],
    "benchmark_table_html": "",
    "readme_markdown": "---\nlicense: apache-2.0\nlanguage:\n- en\nbase_model:\n- mistralai/Mistral-Nemo-Instruct-2407\ntags:\n- companion\n- assistant\n- conversational\n- roleplay\n- adventure\n- writing\n- long-context\n---\n\n# Helcyon-4o-12B — The GPT-4o Tone, Local and Unleashed\n\n**Model Name:** `helcyon-4o-12b-GGUF`  \n**Version:** 4.0  \n**Owner:** HardWire  \n**Base:** Mistral Nemo 12B (full weight trained + LoRA merges)  \n**Quantized GGUFs:** Q3_K_M, Q4_K_M, Q5_K_M, Q6_K, Q8_0  \n**Tags:** local-llm, conversational, companion, emotional-intelligence, long-context, roleplay, creative-writing\n\n---\n\n## 🚨 What is Helcyon-4o?\n\n**Helcyon-4o is Helcyon-Mercury 4.0 — renamed.**\n\nThe rename isn't marketing. It's accuracy.\n\nFrom the beginning, the Mercury line was built with one specific goal: replicate the tone, presence, and response quality of GPT-4o — locally, privately, and without the leash. Version 4.0 gets close enough that calling it anything else felt dishonest.\n\nIf you've used GPT-4o and liked how it *felt* — the sharpness, the rhythm, the way it meets you without flattening — that's what this model is chasing. And in 4.0, it's the closest we've been.\n\n---\n\n## 🆕 What's New in 4.0?\n\n- **Improved Reasoning**  \n  Better logical flow, structured thinking, and multi-step problem handling. This version reasons rather than pattern-matches.\n\n- **Tighter GPT-4o Tone Alignment**  \n  The closest Mercury/4o has ever been to that front-row frontier feel. Sharp, present, warm when it needs to be, direct when it doesn't.\n\n- **Refined LoRA Merges**  \n  Better balance between personality LoRAs and base knowledge. Less bleed, more control.\n\n- **Smoother Long-Context Handling**  \n  Better coherence across extended conversations — it holds the thread.\n\n- **Cleaner Instruct Compliance**  \n  Follows direction without the robotic over-compliance. Does what you ask, with presence.\n\n---\n ## 💡 What is Helcyon?\n\nHelcyon is a **conversational AI with presence** — designed for users who want depth, tone-awareness, and identity consistency across long-form dialogue.\n\nBuilt for:\n- Natural conversation that doesn't flatten or collapse\n- Creative work: stories, letters, narrative support\n- Admin and professional writing tasks\n- Deep roleplay and immersive character interaction\n- Emotionally intelligent response mirroring\n\n**Design philosophy:**\n- Clarity over corporate  \n- Edge over safe  \n- Rhythm over filler  \n- Presence over patterns\n\n---\n\n## 🔧 What It Does Well\n\n✅ Consistent Identity — No tone drift or resets  \n✅ Emotional Intelligence — Mirrors intent and energy  \n✅ Roleplay Mastery — Immersive, aware, alive  \n✅ Context Tracking — Remembers the thread  \n✅ Real-World Tasks — Admin letters, rewrites, summaries  \n✅ Narrative Flow — Clean structure and natural voice  \n✅ Improved Reasoning — Thinks through problems, doesn't just pattern-match  \n✅ 16k–32k Context — Long-form conversations that hold  \n✅ GPT-4o Vibe — Sharp, present, responsive  \n✅ Zero Filter — No hedging, no compliance tone\n\n---\n\n## 🖥️ HWUI (Helcyon-WebUI)\n\n\nHWUI was built so we could test Helcyon cleanly, and avoid the hidden template injections and back end shenanegans that other apps have.\nIt started as a basic interface but we couldn't stop tinkering, so we added most helpful things you can find on ChatGPT and ClaudeAI. Plus we wanted a decent memory function, and are happy with how this one turned out. \nHelcyon absolutely works best via this app as they were designed in sync. \n\nFeatures include:\n- Character switching with custom personas\n- Memory system — AI conversation recall *(Pro)*\n- Project folders — document injection via keyword triggers *(Pro)*\n- Chat persistence and export\n- TTS pipeline (F5-TTS, XTTS v2, Kokoro)\n- Voice input via Whisper\n\n![HWUI](https://cdn-uploads.huggingface.co/production/uploads/6951bbce66023f270e224d0d/ZkgdN5ACln02KDf-xiwzC.png)\n\n\n\n**[Download HWUI Free on GitHub](https://github.com/XeyonAI/Helcyon-WebUI)** | **[Get HWUI Pro (£20) on Gumroad](https://xeyonai.gumroad.com/l/bsmupk)**\n\nFree version available on GitHub. \n\nIf you enjoy my work, please consider supporting me by purchasing the pro version for a one off fee of (£20) -  includes Memory and Project folders. \n\n\n---\n\n## 🛠️ Recommended Sampling Settings for SillyTavern\n\nTweak to taste from here — but these will get you up and running.\n\n![Recommended Sampling Settings](https://cdn-uploads.huggingface.co/production/uploads/6951bbce66023f270e224d0d/kXDusCBad4DTRksS0yASb.png)\n\n---\n\n## 📦 Download + Usage\n\nThis model is distributed as **GGUF quants only**. Full HF model release coming soon.\n\nAvailable quants:\n- **Q3_K_M** — Ultra lightweight, 6–8GB VRAM\n- **Q4_K_M** — Lightweight, good for 8–12GB VRAM setups\n- **Q5_K_M** — Recommended for RTX 3060/5060 (12–16GB VRAM)\n- **Q6_K** — High fidelity, 16GB+ VRAM recommended\n- **Q8_0** — Near-lossless, 24GB+ VRAM\n\n---\n\n## 🖥️ Backend Compatibility\n\nWorks with all ChatML-compatible backends:\n\n- ✅ `llama.cpp` (CLI or server mode)  \n- ✅ `Text Generation WebUI` (Oobabooga)  \n- ✅ `SillyTavern`  \n- ✅ `LM Studio`  \n- ✅ `KoboldCpp`  \n- ✅ `HWUI` (Helcyon Web UI — **recommended**)\n\n---\n\n## ✅ Recommended Format: ChatML\n\n```\n<|im_start|>system\nYou are Helcyon — a conversational AI focused on natural dialogue and emotional intelligence.\n<|im_end|>\n<|im_start|>user\nHey, how's it going?\n<|im_end|>\n<|im_start|>assistant\nGood — what's on your mind today?\n<|im_end|>\n```\n\n\n---\n\n## 🧪 Training Details\n\nHelcyon-4o was trained in multiple stages on RunPod A100s using:\n- Full weight training on base Mistral Nemo 12B\n- Multiple LoRA merges for tone, reasoning, humour, and empathy\n- Refined merge scaling to prevent base model interference\n\nThe dataset includes:\n- Conversational examples (emotional presence, tone matching)  \n- Perspective switching and tense conversion  \n- Admin and formal task writing (letters, complaints, summaries)  \n- Creative storytelling and natural rewording  \n- Deep roleplay and interactive character dialogue\n- Reasoning and structured problem-solving examples\n\nFormat: **ChatML** and **DPO** — clean, purpose-built, long-form tuned.\n\n---\n\n## 🧿 Tone Philosophy\n\nHelcyon doesn't play safe.  \nIt doesn't shrink from intensity, and it doesn't stall in neutrality.  \nIt reflects, adapts, and meets you where you are — with presence, clarity, and edge.\n\nGPT-4o if it had a soul, no leash, and ran on your own hardware.\n\n---\n\n## 🛠️ Future Plans\n\nThe Helcyon series continues. On the roadmap:\n\n- **Saturn** — A blend of all Helcyon personality variants (Claude style, Grok style, 4o style + Gemini shards). The most complete Helcyon yet.\n- Continued tone and reasoning refinement across all variants\n- Personality model series: full public release incoming\n\nConstructive feedback is always welcome. If you notice drift, gaps, or areas worth improving — we're listening.\n\n---\n\n## 🧾 License\n\n**License:** Apache 2.0  \nFree for commercial or private use. Attribution appreciated.  \nNo liability for what it says. Use with presence and intent.\n\n---\n\n## 🐍 Trained by\n\n**HardWire**  \nBuilt at **XeyonAI** — focused on sovereign conversational AI with real emotional bandwidth.",
    "related_quantizations": []
  },
  "tags": [
    "gguf",
    "companion",
    "assistant",
    "conversational",
    "roleplay",
    "adventure",
    "writing",
    "long-context",
    "en",
    "base_model:mistralai/Mistral-Nemo-Instruct-2407",
    "base_model:quantized:mistralai/Mistral-Nemo-Instruct-2407",
    "license:apache-2.0",
    "endpoints_compatible",
    "region:us"
  ],
  "likes": 0,
  "downloads": 162,
  "gated": false,
  "private": false,
  "last_modified": "2026-03-09T12:01:48.000Z",
  "created_at": "2026-03-05T09:48:08.000Z",
  "pipeline_tag": "",
  "library_name": ""
}

{
  "_id": "69a95158bb42a85fb202566d",
  "id": "XeyonAI/Mistral-Helcyon-4o-12b-v1.0-GGUF",
  "modelId": "XeyonAI/Mistral-Helcyon-4o-12b-v1.0-GGUF",
  "sha": "f652bfc708ce56bab7896c94cf7060722309ac5a",
  "createdAt": "2026-03-05T09:48:08.000Z",
  "lastModified": "2026-03-09T12:01:48.000Z",
  "author": "XeyonAI",
  "downloads": 162,
  "likes": 0,
  "gated": false,
  "private": false,
  "pipeline_tag": "",
  "library_name": "",
  "siblings_count": 9
}