{
  "schema": 1,
  "metric": "Client time to first nonempty streamed content, seconds; includes queueing",
  "forty_k": {
    "a": {
      "source_run": "a-40k",
      "resource": {
        "cache_gib": 5,
        "max_model_len": 65536,
        "max_num_seqs": 1
      },
      "cases": [
        {
          "case": "cold",
          "ttft_seconds": 13.984786209010053,
          "elapsed_seconds": 14.68904410302639
        },
        {
          "case": "repeat",
          "ttft_seconds": 14.063400466984604,
          "elapsed_seconds": 14.769380340003408
        },
        {
          "case": "grow1",
          "ttft_seconds": 14.203211338026449,
          "elapsed_seconds": 14.673154334013816
        },
        {
          "case": "grow2",
          "ttft_seconds": 14.387746015039738,
          "elapsed_seconds": 14.860299040039536
        }
      ]
    },
    "c": {
      "source_run": "c-40k",
      "resource": {
        "cache_gib": 5,
        "max_model_len": 65536,
        "max_num_seqs": 1
      },
      "cases": [
        {
          "case": "cold",
          "ttft_seconds": 15.57830987899797,
          "elapsed_seconds": 16.300932092999574
        },
        {
          "case": "repeat",
          "ttft_seconds": 1.133074699959252,
          "elapsed_seconds": 1.857072235958185
        },
        {
          "case": "grow1",
          "ttft_seconds": 1.2307560829794966,
          "elapsed_seconds": 1.7105457639554515
        },
        {
          "case": "grow2",
          "ttft_seconds": 1.3864010990364477,
          "elapsed_seconds": 1.8673484140308574
        }
      ]
    },
    "d": {
      "source_run": "d-40k",
      "resource": {
        "cache_gib": 5,
        "max_model_len": 65536,
        "max_num_seqs": 1
      },
      "cases": [
        {
          "case": "cold",
          "ttft_seconds": 14.679737413011026,
          "elapsed_seconds": 15.40079564001644
        },
        {
          "case": "repeat",
          "ttft_seconds": 1.0891118120052852,
          "elapsed_seconds": 1.8119888529763557
        },
        {
          "case": "grow1",
          "ttft_seconds": 1.1789259089855477,
          "elapsed_seconds": 1.657018878031522
        },
        {
          "case": "grow2",
          "ttft_seconds": 1.3257642930257134,
          "elapsed_seconds": 1.8077886269893497
        }
      ]
    }
  },
  "long_context": {
    "primary": {
      "source_run": "d-150k",
      "resource": {
        "cache_gib": 8,
        "max_model_len": 160000,
        "max_num_seqs": 1
      },
      "cases": [
        {
          "case": "cold",
          "ttft_seconds": 88.45819045801181,
          "elapsed_seconds": 89.47866020002402
        },
        {
          "case": "repeat",
          "ttft_seconds": 2.0438307669828646,
          "elapsed_seconds": 3.0627995359827764
        },
        {
          "case": "grow1",
          "ttft_seconds": 2.1468376129632816,
          "elapsed_seconds": 2.7497615829925053
        },
        {
          "case": "grow2",
          "ttft_seconds": 2.540299961983692,
          "elapsed_seconds": 3.144473886000924
        }
      ]
    },
    "confirmation": {
      "source_run": "d-repeat-150k",
      "resource": {
        "cache_gib": 8,
        "max_model_len": 160000,
        "max_num_seqs": 1
      },
      "cases": [
        {
          "case": "cold",
          "ttft_seconds": 88.61075399402762,
          "elapsed_seconds": 89.57036234997213
        },
        {
          "case": "repeat",
          "ttft_seconds": 2.032992133987136,
          "elapsed_seconds": 2.990930464991834
        }
      ]
    }
  },
  "publication_scope": "Report only. Native-prefill APC candidate is not distributed. Existing image stock-GDN APC uses the documented profile override."
}
