{
  "conversion": {
    "apple_inference_required": true,
    "cuda_version": "0.30.0",
    "linux_conversion_probe": [
      "quantize",
      "dequantize"
    ],
    "quantization": {
      "bits": [
        4,
        5,
        6,
        8
      ],
      "group_size": 64,
      "mode": "affine"
    },
    "source_revision": "a1e0e0b56904cb8eef3a4671b747833c94b0b9fb",
    "version": "0.30.0"
  },
  "inference": {
    "llama_cpp_build": "b10034",
    "mlx_c_revision": "fba4470b8907",
    "mlx_version": "0.32.0",
    "off_mode": "MTP weights loaded, speculative decoding disabled",
    "on_mode": "MTP weights loaded and draft tokens observed",
    "revision": "mlx-serve-26.8.7|mlx-0.32.0|mlx-c-fba4470b8907|llama.cpp-b10034",
    "version": "26.8.7"
  },
  "required_gates": [
    "M4 Pro MTP-off load/generate",
    "M4 Pro MTP-on load/generate with draft tokens"
  ],
  "runtime_evidence_pending_at_packaging": true,
  "schema": 2
}