{
  "source": "Tempo traces from the maintainer\u2019s managoat.com instance",
  "window_start_utc": "2026-10-15T12:58:33+00:00",
  "window_end_utc": "2026-10-16T12:58:33+00:00",
  "query": "{resource.service.name=\"fountain\" && name=\"fountain.turn\" && span.response_path=\"cold\"} | select(span.provider, span.runtime, span.response_path, span.response_first_text_ms, span.effective_model, span.response_phase_admission_queue_ms, span.response_phase_sandbox_ms, span.response_phase_bootstrap_ms, span.response_phase_runtime_session_ms, span.response_phase_prompt_text_ms)",
  "matched_cold_turns": 130,
  "complete_phase_samples": 80,
  "measurement": "Cold means a new conversation launched with a prompt. Sandbox + setup sums the exclusive sandbox provisioning and bootstrap phases per request, then takes the median. Request to first text measures application entry to the first nonempty assistant-text chunk, including queueing, provisioning, setup, runtime startup and agent/model work, excluding client transport and rendering.",
  "limitations": [
    "Only traced cold turns that reached first text and carried every phase are included. Missing-phase traces are excluded from both displayed columns. These samples do not measure failed launches or establish a success rate.",
    "Observed snapshot, not a controlled benchmark or latency guarantee. Models and workloads vary.",
    "Sprites/Codex has only two samples.",
    "130 traces returned below the 500-trace cap; all 27 search jobs completed."
  ],
  "results": [
    {
      "provider": "Sprites",
      "runtime": "Claude Code",
      "samples": 35,
      "models": {
        "opus": 1,
        "haiku": 31,
        "sonnet": 3
      },
      "sandbox_setup_median_seconds": 4.685,
      "first_text_median_seconds": 8.561,
      "sandbox_setup_seconds": [
        4.947,
        4.886,
        4.715,
        4.522,
        4.027,
        4.068,
        4.478,
        4.634,
        4.925,
        4.999,
        4.582,
        4.268,
        4.621,
        4.9,
        4.281,
        5.017,
        4.685,
        4.654,
        4.741,
        5.093,
        4.47,
        4.551,
        4.196,
        4.557,
        4.818,
        4.051,
        4.027,
        4.483,
        4.792,
        6.26,
        5.16,
        4.992,
        6.023,
        5.778,
        6.179
      ],
      "request_first_text_seconds": [
        142.365,
        8.812,
        7.742,
        8.584,
        7.134,
        7.632,
        8.816,
        8.355,
        8.521,
        8.993,
        7.86,
        7.5,
        8.227,
        8.466,
        7.448,
        9.308,
        7.708,
        9.886,
        9.149,
        9.492,
        8.228,
        8.944,
        7.563,
        8.983,
        8.561,
        7.4,
        7.199,
        7.919,
        9.163,
        11.048,
        8.145,
        9.096,
        11.039,
        9.671,
        9.985
      ]
    },
    {
      "provider": "E2B",
      "runtime": "Claude Code",
      "samples": 36,
      "models": {
        "haiku": 33,
        "sonnet": 3
      },
      "sandbox_setup_median_seconds": 5.039,
      "first_text_median_seconds": 8.0165,
      "sandbox_setup_seconds": [
        8.962,
        8.424,
        7.66,
        5.534,
        5.007,
        4.907,
        4.913,
        5.033,
        5.013,
        4.978,
        4.662,
        4.954,
        8.555,
        4.996,
        4.934,
        4.913,
        5.079,
        5.034,
        5.247,
        5.345,
        4.769,
        5.077,
        5.044,
        5.079,
        4.808,
        5.49,
        5.328,
        4.685,
        5.131,
        4.973,
        4.886,
        4.916,
        5.641,
        5.848,
        8.954,
        7.274
      ],
      "request_first_text_seconds": [
        11.663,
        11.281,
        10.76,
        8.496,
        7.868,
        8.014,
        7.527,
        8.147,
        8.019,
        7.477,
        7.736,
        7.416,
        10.79,
        8.043,
        7.913,
        7.406,
        7.78,
        7.624,
        9.091,
        8.143,
        7.691,
        8.428,
        7.984,
        7.581,
        7.445,
        8.481,
        8.12,
        7.238,
        8.036,
        7.427,
        7.96,
        7.634,
        8.631,
        8.591,
        11.878,
        10.58
      ]
    },
    {
      "provider": "Sprites",
      "runtime": "Codex",
      "samples": 2,
      "models": {
        "gpt-6-astra": 2
      },
      "sandbox_setup_median_seconds": 9.8375,
      "first_text_median_seconds": 20.676499999999997,
      "sandbox_setup_seconds": [
        9.496,
        10.179
      ],
      "request_first_text_seconds": [
        15.883,
        25.47
      ]
    },
    {
      "provider": "E2B",
      "runtime": "Codex",
      "samples": 7,
      "models": {
        "gpt-6-astra": 7
      },
      "sandbox_setup_median_seconds": 9.182,
      "first_text_median_seconds": 13.899,
      "sandbox_setup_seconds": [
        9.851,
        9.182,
        8.717,
        9.765,
        6.227,
        6.913,
        9.727
      ],
      "request_first_text_seconds": [
        15.809,
        14.171,
        13.287,
        13.899,
        11.936,
        12.786,
        17.302
      ]
    }
  ]
}
