{
  "schema_version": 1,
  "model": "Singularity xMESH 0.8B",
  "artifact": "Singularity-xMESH-0.8B-BF16.gguf",
  "artifact_sha256": "9f4c08f19c4d203059bc74ed2754e800ec472d7276d8ddb2321480e86d392fff",
  "llama_cpp": {
    "tag": "b11719",
    "commit": "5aeac37a35b44fde4bdffaec9159f2fcb253b622"
  },
  "profiles": {
    "macos_arm64_cpu": {
      "status": "passed",
      "cases": 80,
      "matching_decisions": 960,
      "total_decisions": 960,
      "median_request_ms": 3109.9937505,
      "p95_request_ms": 3495.010333,
      "slots": 1
    },
    "macos_arm64_metal": {
      "status": "passed",
      "gpu_layers": 99,
      "cases": 80,
      "matching_decisions": 959,
      "total_decisions": 960,
      "note": "BF16 full Metal offload matches the CPU reference except one near-tie decision. See ACCELERATOR_PARITY.json."
    },
    "linux_arm64_cpu": {
      "status": "passed",
      "environment": "Docker Desktop aarch64, Ubuntu 24.04, libgomp1",
      "archive": "llama-b11719-bin-ubuntu-arm64.tar.gz",
      "archive_sha256": "290aa4432ad8e28b638285be6a75d0dd6f327d9b2d95f7048a8cf5e69704bee9",
      "server_sha256": "7b0734c19cc38e09d7c9b4ed96cbdbb54f9318282cecd7e4a82f067b54451892",
      "fixture_input_tokens": 1012,
      "expected_decisions": 9,
      "matching_decisions": 9,
      "response_sha256": "b1d70e2b94d1157f78b840d7ba1372e5719ca34f5231920fc65dfe27dd50bf20",
      "dependency": "libgomp.so.1 from Ubuntu package libgomp1"
    },
    "linux_x64_cpu": {
      "status": "passed",
      "environment": "GitHub Actions ubuntu-24.04 x64, Vulkan distribution with -ngl 0, libgomp1 and libvulkan1",
      "workflow_run": "Singularity-Layer/sgl-network-node/actions/runs/39158899243",
      "archive": "llama-b11719-bin-ubuntu-vulkan-x64.tar.gz",
      "archive_sha256": "9a09cce245acf8823feb99fb7ba8ddd3e878e3827564888131e0185044aa4429",
      "gpu_layers": 0,
      "fixture_input_tokens": 1012,
      "expected_decisions": 9,
      "matching_decisions": 9,
      "latency_ms": 17155.671278
    },
    "windows_x64_cpu": {
      "status": "passed",
      "environment": "GitHub Actions windows-2025-vs2026 x64, Vulkan distribution with -ngl 0",
      "workflow_run": "Singularity-Layer/sgl-network-node/actions/runs/39158899243",
      "archive": "llama-b11719-bin-win-vulkan-x64.zip",
      "archive_sha256": "08dde7f4133443ac629431c35f0435e267ca73251fc002cb95e10625424e2963",
      "gpu_layers": 0,
      "fixture_input_tokens": 1012,
      "expected_decisions": 9,
      "matching_decisions": 9,
      "latency_ms": 10955.597100000006
    },
    "linux_x64_nvidia_vulkan": {
      "status": "passed",
      "environment": "NVIDIA RTX A4000, Ubuntu 22.04 container, driver 550.144.03",
      "archive": "llama-b11719-bin-ubuntu-vulkan-x64.tar.gz",
      "archive_sha256": "9a09cce245acf8823feb99fb7ba8ddd3e878e3827564888131e0185044aa4429",
      "gpu_layers": 99,
      "cases": 80,
      "matching_decisions": 960,
      "total_decisions": 960,
      "median_request_ms": 305.5,
      "note": "See ACCELERATOR_PARITY.json."
    }
  }
}