{
  "schema": "fak-hardware-latest/1",
  "as_of": "2026-09-06",
  "platforms": {
    "Mac": {
      "observed": "2026-09-03",
      "detail": "docs/notes/MAC-THREEWAY-BENCH-2026-09-03.md",
      "row": "| Mac | Qwen3.8-27B Q4_K_M on an Apple M3 Pro: 7.61 decode tok/s (+3.1% vs llama.cpp 7.38, MLX 8.07) and 12.6 ms prefix TTFT, observed 2026-09-03. | Verified matched-envelope single-stream decode leads llama.cpp Metal; RadixAttention prefix caching eliminates repeat prefill. | [Mac result](docs/notes/MAC-THREEWAY-BENCH-2026-09-03.md) |"
    },
    "AMD": {
      "observed": "2026-06-19",
      "detail": "docs/benchmarks/QWEN36-AMD-VULKAN-RESULTS.md",
      "row": "| AMD | Qwen3.6-27B on an RX 7600: the measured pure-fak microbench reached 1.15–1.24 decode tok/s versus 0.99 for the local llama.cpp Vulkan baseline, observed 2026-06-19. | Witnessed in that narrow microbench; not a broad quality or full-model parity claim. Qwen3.8 awaits a comparable AMD receipt. | [AMD result](docs/benchmarks/QWEN36-AMD-VULKAN-RESULTS.md) |"
    },
    "NVIDIA": {
      "observed": "2026-09-05",
      "detail": "docs/_witnesses/issue-10944-nvidia-gcp-overnight/README.md",
      "row": "| NVIDIA | Hopper H100 Q8_0 decode reached 111.9 tok/s (+17.4% vs f32); live A100 Qwen3.8-27B prefix reuse achieved 4.84× TTFT speedup, observed 2026-09-05. | Witnessed on physical GCP H100 (a3-highgpu-1g) & A100; matched Q8 device GEMV and 50-agent concurrency grid (91/91 ok). | [NVIDIA result](docs/_witnesses/issue-10944-nvidia-gcp-overnight/README.md) |"
    }
  }
}