{
  "schema_version": "2.0.0",
  "benchmark": "teanga-bench",
  "benchmark_name": "Teanga Bench",
  "benchmark_version": "0.2.0",
  "suite": "tonada",
  "generated_at": "2026-11-15T23:00:41+00:00",
  "note": "Only live runs with status complete are ranked. Scores are means over items with a 95% bootstrap interval; the suite score weights every test equally. cost_usd is the spend on this suite's tests. by_l1 gives the mean of per-test scores.",
  "models": [
    {
      "rank": 1,
      "ranked": true,
      "model": "openai/gpt-6-luna",
      "family": "openai",
      "page": "models/openai-gpt-6-luna.html",
      "run_file": "openai__gpt-6-luna.json",
      "run_kind": "live",
      "status": "complete",
      "generated_at": "2026-11-15T23:00:41+00:00",
      "data": "full",
      "epochs": 3,
      "overall": {
        "score": 0.84,
        "ci95": [
          0.8161111111111111,
          0.8630590277777778
        ],
        "n_tests": 2,
        "n_items": 760
      },
      "cost_usd": 0.808824467,
      "cost_per_correct_usd": 0.0004186462044513458,
      "by_l1": {
        "japanese": 0.8455555555555556,
        "portuguese": 0.8455555555555556,
        "spanish": 0.8325925925925926,
        "vietnamese": 0.8362962962962963
      },
      "tests": {
        "exercises": {
          "score": 0.98,
          "ci95": [
            0.965,
            0.9925
          ],
          "n_items": 400,
          "epochs": 3,
          "consistency": 1.0,
          "status": "complete",
          "dataset": "full",
          "cost_usd": 0.0333021,
          "cost_per_correct_usd": 2.831811224489796e-05
        },
        "tutor_accuracy": {
          "score": 0.7,
          "ci95": [
            0.6527777777777778,
            0.7435185185185185
          ],
          "n_items": 360,
          "epochs": 3,
          "consistency": 0.8555555555555555,
          "status": "complete",
          "dataset": "full",
          "cost_usd": 0.775522367,
          "cost_per_correct_usd": 0.001025823236772487
        }
      }
    }
  ]
}