{
  "scenario_id": "informes_7b_q4",
  "model": {
    "parameters": 7000000000,
    "weight_bits": 4,
    "layers": 32,
    "kv_heads": 8,
    "head_dim": 128,
    "kv_bytes": 2
  },
  "serving": {
    "batch": 16,
    "context_tokens": 2048,
    "output_tokens_per_user": 1200,
    "aggregate_decode_tokens_per_second": 240,
    "runtime_margin_gb": 6
  }
}