kuroto4ka/qwen3-vl-8b-instruct-unredacted-max-quants-gguf overview

{
  "metadata": {},
  "card_data": {
    "license": "apache-2.0",
    "language": [
      "en",
      "ru",
      "zh"
    ],
    "tags": [
      "vision",
      "multimodal",
      "gguf",
      "qwen",
      "qwen-3",
      "unredacted",
      "image-to-text",
      "text-generation",
      "conversational",
      "roleplay",
      "assistant",
      "vl",
      "vlm"
    ],
    "pipeline_tag": "image-to-text",
    "library_name": "gguf",
    "base_model": [
      "prithivMLmods/Qwen3-VL-8B-Instruct-Unredacted-MAX"
    ],
    "frontmatter": {
      "license": "apache-2.0",
      "language": [
        "en",
        "ru",
        "zh"
      ],
      "tags": [
        "vision",
        "multimodal",
        "gguf",
        "qwen",
        "qwen-3",
        "unredacted",
        "image-to-text",
        "text-generation",
        "conversational",
        "roleplay",
        "assistant",
        "vl",
        "vlm"
      ],
      "pipeline_tag": "image-to-text",
      "library_name": "gguf",
      "base_model": [
        "prithivMLmods/Qwen3-VL-8B-Instruct-Unredacted-MAX"
      ]
    },
    "hero_image_url": "https://camo.githubusercontent.com/17b4379eedbf639f0fc005e6512f2a629b9cb059d3cc4eacc1ec65fa21f92898/68747470733a2f2f7169616e77656e2d7265732e6f73732d616363656c65726174652e616c6979756e63732e636f6d2f5177656e332d564c2f7177656e33766c6c6f676f2e706e67",
    "summary": "This repository contains high-quality GGUF quantizations for the prithivMLmods/Qwen3-VL-8B-Instruct-Unredacted-MAX model.",
    "quick_links": [],
    "benchmark_table_html": "",
    "readme_markdown": "---\nlicense: apache-2.0\nlanguage:\n- en\n- ru\n- zh\ntags:\n- vision\n- multimodal\n- gguf\n- qwen\n- qwen-3\n- unredacted\n- image-to-text\n- text-generation\n- conversational\n- roleplay\n- assistant\n- vl\n- vlm\npipeline_tag: image-to-text\nlibrary_name: gguf\nbase_model:\n- prithivMLmods/Qwen3-VL-8B-Instruct-Unredacted-MAX\n---\n\n![1](https://camo.githubusercontent.com/17b4379eedbf639f0fc005e6512f2a629b9cb059d3cc4eacc1ec65fa21f92898/68747470733a2f2f7169616e77656e2d7265732e6f73732d616363656c65726174652e616c6979756e63732e636f6d2f5177656e332d564c2f7177656e33766c6c6f676f2e706e67)\n\n# Qwen3-VL-8B-Instruct-Unredacted-MAX-Quants-GGUF\n\nThis repository contains high-quality GGUF quantizations for the [prithivMLmods/Qwen3-VL-8B-Instruct-Unredacted-MAX](https://huggingface.co/prithivMLmods/Qwen3-VL-8B-Instruct-Unredacted-MAX) model.\n\n##  Highlights\n- **Unredacted & MAX**: Maximum performance version without restrictive filters.\n- **Full Vision Support**: Includes multiple versions of the vision projector (`mmproj`) for different hardware needs.\n- **Optimized**: Compatible with the latest `llama.cpp` and other GGUF-supported backends.\n\n##  Files Included\n\n### 1. Model Weights (LLM)\n\n| Filename | Quant Method | Description |\n| :--- | :--- | :--- |\n| `Q4_K_M.gguf` | Q4_K_M | **Recommended.** Best balance of speed and intelligence. |\n| `Q8_0.gguf` | Q8_0 | High quality, nearly identical to original weights. |\n| `Q6_K.gguf` | Q6_K | Very high quality, slightly slower than Q4. |\n| `Q5_K_M.gguf` | Q5_K_M | Good balance between Q4 and Q6. |\n| `Q3_K_M.gguf` | Q3_K_M | Low size, moderate quality loss. |\n| `Q2_K.gguf` | Q2_K | Smallest possible size, significant quality loss. |\n| `F16.gguf` | F16 | Baseline reference quality. |\n\n### 2. Vision Projectors (mmproj)\n*Required for image recognition tasks.*\n\n\n| Filename | Type | Description |\n| :--- | :--- | :--- |\n| `mmproj-f32.gguf` | F32 | Absolute maximum precision (2.3GB). |\n| `mmproj-f16.gguf` | F16 | Industry standard for high-quality vision. |\n| `mmproj-bf16.gguf` | BF16 | Optimized for modern NVIDIA GPUs (Ampere+). |\n| `mmproj-q8_0.gguf` | Q8_0 | Best for saving VRAM without losing recognition detail. |\n\n##  Usage\nTo use vision capabilities in `llama.cpp`, use the following command:\n\n```bash\n./llama-cli -m Qwen3-VL-8B-Instruct-Unredacted-MAX.Q4_K_M.gguf \\\n            --mmproj Qwen3-VL-8B-Instruct-Unredacted-MAX.mmproj-f16.gguf \\\n            --image path/to/your/image.jpg \\\n            -p \"Describe this image\"\n",
    "related_quantizations": []
  },
  "tags": [
    "gguf",
    "vision",
    "multimodal",
    "qwen",
    "qwen-3",
    "unredacted",
    "image-to-text",
    "text-generation",
    "conversational",
    "roleplay",
    "assistant",
    "vl",
    "vlm",
    "en",
    "ru",
    "zh",
    "base_model:prithivMLmods/Qwen3-VL-8B-Instruct-Unredacted-MAX",
    "base_model:quantized:prithivMLmods/Qwen3-VL-8B-Instruct-Unredacted-MAX",
    "license:apache-2.0",
    "endpoints_compatible",
    "region:us"
  ],
  "likes": 1,
  "downloads": 1897,
  "gated": false,
  "private": false,
  "last_modified": "2026-04-13T09:23:21.000Z",
  "created_at": "2026-04-10T21:23:25.000Z",
  "pipeline_tag": "image-to-text",
  "library_name": "gguf"
}

{
  "_id": "69d96a4d44072dd46eed610d",
  "id": "KuroTo4ka/Qwen3-VL-8B-Instruct-Unredacted-MAX-Quants-GGUF",
  "modelId": "KuroTo4ka/Qwen3-VL-8B-Instruct-Unredacted-MAX-Quants-GGUF",
  "sha": "d2db200bba5cc65c50959a5d5624e34d6cc8e7bc",
  "createdAt": "2026-04-10T21:23:25.000Z",
  "lastModified": "2026-04-13T09:23:21.000Z",
  "author": "KuroTo4ka",
  "downloads": 1897,
  "likes": 1,
  "gated": false,
  "private": false,
  "pipeline_tag": "image-to-text",
  "library_name": "gguf",
  "siblings_count": 16
}