{
  "generatedAt": "2026-05-10T06:08:00Z",
  "taskId": "da30aed8-0525-4c28-83e6-65c4cdca43d3",
  "hardware": {
    "gpu": "NVIDIA GeForce RTX 4090 Laptop GPU",
    "gpuMemoryMiB": 16376,
    "driver": "591.74",
    "cudaReportedByNvidiaSmi": "13.1",
    "cpu": "13th Gen Intel(R) Core(TM) i9-13980HX",
    "runtime": "llama.cpp build ff4affb4c / 8067",
    "backend": "Vulkan",
    "device": "Vulkan1"
  },
  "assumptions": {
    "electricityUsdPerKwh": 0.2,
    "conservativeWallPowerWatts": 250,
    "gpuPowerCapWatts": 155,
    "hardwareAmortizationUsdPerHour": 0.342,
    "hardwareAmortizationNote": "Example only: $2000 GPU/system value over 2 years at 8 hours/day. If the hardware is already owned and idle, marginal hardware amortization is $0."
  },
  "benchmarks": [
    {
      "model": "Qwen3-1.7B-Q8_0.gguf",
      "modelType": "qwen3 1.7B Q8_0",
      "modelSizeBytes": 1828474880,
      "parameters": 1720574976,
      "command": "llama-bench -m Qwen3-1.7B-Q8_0.gguf -dev Vulkan1 -ngl 99 -p 512 -n 128 -r 3 -o json",
      "promptProcessingTokensPerSecond": 16757.490604,
      "generationTokensPerSecond": 225.733282,
      "generatedTokensPerHour": 812639.8152,
      "energyOnlyUsdPerMillionGeneratedTokensAt250W": 0.061527,
      "energyOnlyUsdPerMillionGeneratedTokensAt155W": 0.038147,
      "amortizedUsdPerMillionGeneratedTokensAt250W": 0.482379
    },
    {
      "model": "Qwen3-4B-Instruct-2507-Q4_K_M.gguf",
      "modelType": "qwen3 4B Q4_K - Medium",
      "modelSizeBytes": 2491323904,
      "parameters": 4022468096,
      "command": "llama-bench -m Qwen3-4B-Instruct-2507-Q4_K_M.gguf -dev Vulkan1 -ngl 99 -p 512 -n 128 -r 3 -o json",
      "promptProcessingTokensPerSecond": 7384.773071,
      "generationTokensPerSecond": 158.350202,
      "generatedTokensPerHour": 570060.7272,
      "energyOnlyUsdPerMillionGeneratedTokensAt250W": 0.08771,
      "energyOnlyUsdPerMillionGeneratedTokensAt155W": 0.05438,
      "amortizedUsdPerMillionGeneratedTokensAt250W": 0.68765
    }
  ],
  "cloudComparison": {
    "source": "https://openai.com/api/pricing/",
    "pricesUsdPerMillionTokens": {
      "gpt-5.5": { "input": 5.0, "output": 30.0 },
      "gpt-5.4": { "input": 2.5, "output": 15.0 },
      "gpt-5.4-mini": { "input": 0.75, "output": 4.5 },
      "gpt-5-mini": { "input": 0.25, "output": 2.0 },
      "gpt-5-nano": { "input": 0.05, "output": 0.4 }
    }
  }
}