[
 {
  "id": "C01",
  "http_status": 200,
  "recorded_at": "2026-11-03T23:10:40+00:00",
  "ms": 0,
  "reply": {
   "slug": "edge-optimizer",
   "title": "Pick a quantization level for an edge device",
   "status": "ok",
   "input": null,
   "total_ms": 44,
   "steps": [
    {
     "call": "POST /recommend",
     "request": {
      "model_size_mb": 7000,
      "device_memory_mb": 4096,
      "baseline_latency_ms": 900,
      "target_latency_ms": 300
     },
     "status": 200,
     "ms": 24,
     "reply": {
      "recommended_quantization": "int4",
      "feasible": true,
      "estimated_size_mb": 875.0,
      "estimated_latency_ms": 225.0,
      "estimated_accuracy_loss_pct": 4.0,
      "device_memory_budget_mb": 3276.8,
      "inputs": {
       "model_size_mb": 7000.0,
       "device_memory_mb": 4096.0,
       "baseline_latency_ms": 900.0,
       "target_latency_ms": 300.0
      },
      "note": "Estimates are heuristics from a static assumption table (compression/latency factors per quantization level), not measured benchmarks of this specific model."
     }
    }
   ],
   "note": "Real calls to the running service with the public demo identity; replies are unedited."
  }
 },
 {
  "id": "C02",
  "http_status": 200,
  "recorded_at": "2026-11-03T23:10:40+00:00",
  "ms": 0,
  "reply": {
   "slug": "edge-optimizer",
   "title": "Pick a quantization level for an edge device",
   "status": "ok",
   "input": null,
   "total_ms": 44,
   "steps": [
    {
     "call": "POST /recommend",
     "request": {
      "model_size_mb": 7000,
      "device_memory_mb": 4096,
      "baseline_latency_ms": 900,
      "target_latency_ms": 300
     },
     "status": 200,
     "ms": 24,
     "reply": {
      "recommended_quantization": "int4",
      "feasible": true,
      "estimated_size_mb": 875.0,
      "estimated_latency_ms": 225.0,
      "estimated_accuracy_loss_pct": 4.0,
      "device_memory_budget_mb": 3276.8,
      "inputs": {
       "model_size_mb": 7000.0,
       "device_memory_mb": 4096.0,
       "baseline_latency_ms": 900.0,
       "target_latency_ms": 300.0
      },
      "note": "Estimates are heuristics from a static assumption table (compression/latency factors per quantization level), not measured benchmarks of this specific model."
     }
    }
   ],
   "note": "Real calls to the running service with the public demo identity; replies are unedited."
  }
 }
]