{
  "schemaVersion": 1,
  "brand": "Nexi Air",
  "evaluatedOn": "2026-09-16",
  "seed": 20260916,
  "suites": [
    {
      "id": "humaneval",
      "name": "HumanEval",
      "total": 164,
      "correct": 154,
      "completeCorrect": 154,
      "truncated": 0,
      "note": "164 problemas; pass@1, testes Python isolados."
    },
    {
      "id": "ifeval",
      "name": "IFEval estrito",
      "total": 541,
      "correct": 450,
      "completeCorrect": 441,
      "truncated": 14,
      "note": "541 pedidos; avaliador oficial Google, prompts canónicos."
    },
    {
      "id": "gsm8k",
      "name": "GSM8K",
      "total": 200,
      "correct": 194,
      "completeCorrect": 194,
      "truncated": 1,
      "note": "Amostra de 200 dos 1319 problemas."
    },
    {
      "id": "mmlu-pro",
      "name": "MMLU-Pro",
      "total": 200,
      "correct": 168,
      "completeCorrect": 163,
      "truncated": 14,
      "note": "Amostra de 200 dos 12032 problemas, abrangendo 14 áreas."
    },
    {
      "id": "aime-2025",
      "name": "AIME 2025",
      "total": 30,
      "correct": 15,
      "completeCorrect": 15,
      "truncated": 13,
      "note": "30 problemas; raciocínio limitado a 2048 tokens."
    },
    {
      "id": "nexibench-portuguese",
      "name": "Português europeu",
      "total": 121,
      "correct": 101,
      "completeCorrect": 101,
      "truncated": 0,
      "note": "121 tarefas internas; avaliador estrito com falsos negativos conhecidos."
    },
    {
      "id": "contexto-longo",
      "name": "Recuperação em contexto",
      "total": 12,
      "correct": 12,
      "completeCorrect": 12,
      "truncated": 0,
      "note": "12 tarefas, 7448–86097 tokens de entrada; não mede compreensão geral de documentos."
    }
  ],
  "performance": [
    {
      "concurrency": 1,
      "requests": 3,
      "aggregateTokensPerSecond": 29.702,
      "firstAnswerP50Ms": 2095.369,
      "firstAnswerP95Ms": 2116.678
    },
    {
      "concurrency": 2,
      "requests": 6,
      "aggregateTokensPerSecond": 50.599,
      "firstAnswerP50Ms": 2172.048,
      "firstAnswerP95Ms": 3963.415
    },
    {
      "concurrency": 4,
      "requests": 12,
      "aggregateTokensPerSecond": 74.878,
      "firstAnswerP50Ms": 3981.91,
      "firstAnswerP95Ms": 7482.153
    }
  ],
  "coldFirstAnswerMs": 344512.952115,
  "coldUnder25Seconds": false,
  "context": {
    "inputTokens": 260065,
    "firstAnswerMs": 361919.930365,
    "correct": true
  },
  "evidenceSha256": "8639084c621372cfee091cb50032365b26c8a91245380b48fadb1f7a0f22678d"
}