mudler/step-3.5-flash-apex-gguf overview

{
  "metadata": {},
  "card_data": {
    "license": "other",
    "base_model": "stepfun-ai/Step-3.5-Flash",
    "tags": [
      "gguf",
      "quantized",
      "apex",
      "moe",
      "mixture-of-experts",
      "step"
    ],
    "frontmatter": {
      "license": "other",
      "base_model": "stepfun-ai/Step-3.5-Flash",
      "tags": [
        "gguf",
        "quantized",
        "apex",
        "moe",
        "mixture-of-experts",
        "step"
      ]
    },
    "hero_image_url": "",
    "summary": "**APEX (Adaptive Precision for EXpert Models)** quantizations of Step-3.5-Flash. **Brought to you by the LocalAI team** | APEX Project | Technical Report",
    "quick_links": [],
    "benchmark_table_html": "",
    "readme_markdown": "---\nlicense: other\nbase_model: stepfun-ai/Step-3.5-Flash\ntags:\n  - gguf\n  - quantized\n  - apex\n  - moe\n  - mixture-of-experts\n  - step\n---\n\n# Step-3.5-Flash APEX GGUF\n\n**APEX (Adaptive Precision for EXpert Models)** quantizations of [Step-3.5-Flash](https://huggingface.co/stepfun-ai/Step-3.5-Flash).\n\n**Brought to you by the [LocalAI](https://github.com/mudler/LocalAI) team** | [APEX Project](https://github.com/mudler/apex-quant) | [Technical Report](https://github.com/mudler/apex-quant/blob/main/paper/APEX_Technical_Report.pdf)\n\n## Benchmark Results\n\nBenchmarks coming soon. For reference APEX benchmarks on the Qwen3.5-35B-A3B architecture, see [mudler/Qwen3.5-35B-A3B-APEX-GGUF](https://huggingface.co/mudler/Qwen3.5-35B-A3B-APEX-GGUF).\n\n## What is APEX?\n\nAPEX is a quantization strategy for Mixture-of-Experts (MoE) models. It classifies tensors by role (routed expert, shared expert, attention) and applies a layer-wise precision gradient -- edge layers get higher precision, middle layers get more aggressive compression. I-variants use diverse imatrix calibration (chat, code, reasoning, tool-calling, agentic traces, Wikipedia).\n\nSee the [APEX project](https://github.com/mudler/apex-quant) for full details, technical report, and scripts.\n\n## Architecture\n\n- **Model**: Step-3.5-Flash (Step3p5)\n- **Layers**: 45 (3 dense + 42 MoE)\n- **Experts**: 288 routed + 1 shared (top-8 active per token)\n- **Total Parameters**: ~196B\n- **Active Parameters**: ~11B per token\n- **Context**: 256K tokens (MTP 4-token speculative head)\n- **APEX Config**: 5+5 symmetric edge gradient across 45 layers\n\n## Run with LocalAI\n\n```bash\nlocal-ai run mudler/Step-3.5-Flash-APEX-GGUF@Step-3.5-Flash-APEX-I-Balanced.gguf\n```\n\n## Credits\n\nAPEX is brought to you by the [LocalAI](https://github.com/mudler/LocalAI) team. Developed through human-driven, AI-assisted research. Built on [llama.cpp](https://github.com/ggerganov/llama.cpp).\n",
    "related_quantizations": []
  },
  "tags": [
    "gguf",
    "quantized",
    "apex",
    "moe",
    "mixture-of-experts",
    "step",
    "base_model:stepfun-ai/Step-3.5-Flash",
    "base_model:quantized:stepfun-ai/Step-3.5-Flash",
    "license:other",
    "endpoints_compatible",
    "region:us",
    "conversational"
  ],
  "likes": 1,
  "downloads": 1355,
  "gated": false,
  "private": false,
  "last_modified": "2026-04-07T03:23:14.000Z",
  "created_at": "2026-04-07T02:34:35.000Z",
  "pipeline_tag": "",
  "library_name": ""
}

{
  "_id": "69d46d3bf821065224281aac",
  "id": "mudler/Step-3.5-Flash-APEX-GGUF",
  "modelId": "mudler/Step-3.5-Flash-APEX-GGUF",
  "sha": "94d83a1412648a4d7b52bca8abe802b8f5e5622b",
  "createdAt": "2026-04-07T02:34:35.000Z",
  "lastModified": "2026-04-07T03:23:14.000Z",
  "author": "mudler",
  "downloads": 1355,
  "likes": 1,
  "gated": false,
  "private": false,
  "pipeline_tag": "",
  "library_name": "",
  "siblings_count": 9
}