{
  "$comment": "Machine-readable mirror of /facts.html. Must never contain a claim absent from the human-visible page. Keep in sync with site.config.json.",
  "name": "LlamaBox",
  "description": "An Android application that runs compatible GGUF language models on-device using llama.cpp. After a model is downloaded or imported, chat generation and local history do not require a network connection.",
  "canonicalUrl": "https://llamabox-ai.vercel.app/",
  "lastVerified": "2026-09-16",
  "platform": {
    "os": "Android",
    "minimumVersion": "7.0",
    "minimumApiLevel": 24,
    "targetApiLevel": 36,
    "architecture": "arm64-v8a",
    "iosAvailable": false
  },
  "package": "com.llamabox",
  "versionName": "1.0.7",
  "engine": {
    "inference": "llama.cpp via llama.rn",
    "modelFormat": "GGUF",
    "recommendedQuantization": "Q4_K_M",
    "compute": "CPU by default; experimental acceleration may be available on supported devices",
    "gpuAcceleration": {
      "status": "experimental",
      "enabledByDefault": false,
      "availability": "device and build dependent",
      "fallback": "CPU"
    },
    "npuAcceleration": false,
    "defaultContextLength": 2048,
    "visionContextLength": 4096,
    "visionRequiresProjector": true
  },
  "release": {
    "stage": "public-release",
    "label": "Available on Google Play",
    "channel": "Google Play public release",
    "publicDownloadAvailable": true,
    "productionApproved": true,
    "publicReleaseDate": null,
    "$comment": "Public availability confirmed 2026-09-07. Exact release date is not recorded; source availability is separate.",
    "downloadUrl": "https://play.google.com/store/apps/details?id=com.llamabox"
  },
  "privacy": {
    "accountRequiredForLocalChat": false,
    "conversationStorage": "Local SQLite in app-private storage on the device",
    "uploadsConversations": false,
    "uploadsPrompts": false,
    "uploadsGeneratedResponses": false,
    "analyticsSdk": false,
    "crashReportingSdk": false,
    "inAppUpdateCheck": false,
    "outboundHosts": [
      "huggingface.co"
    ],
    "outboundHostsPurpose": "Model discovery and downloads initiated from the app.",
    "llamaboxInferenceServer": false,
    "localApiServer": {
      "present": true,
      "enabledByDefault": true,
      "scope": "Can become active when local models are available and serves compatible clients on the user's local network. Inference remains on-device."
    },
    "declaredAndroidPermissions": [
      "android.permission.INTERNET",
      "android.permission.READ_EXTERNAL_STORAGE",
      "android.permission.WRITE_EXTERNAL_STORAGE",
      "android.permission.RECORD_AUDIO",
      "android.permission.SYSTEM_ALERT_WINDOW",
      "android.permission.VIBRATE"
    ],
    "textToSpeech": "Uses the Android system TTS engine selected by the user; that engine's own network behaviour applies.",
    "speechToText": "Not included in the current public capability set.",
    "verificationBasis": "Verified by reading the application dependency manifest and source tree. NOT independently auditable by third parties yet, because the app source is not public.",
    "policyUrl": "https://llamabox-ai.vercel.app/privacy.html",
    "networkBehaviorUrl": "https://llamabox-ai.vercel.app/network-behavior.html"
  },
  "licensing": {
    "appCost": "No charge",
    "sourcePublic": false,
    "plannedSourceLicense": "AGPL-3.0",
    "commercialLicenseAvailable": true,
    "trademarkReserved": true,
    "modelLicenses": "Models are published by third parties under their own licenses, some restricting commercial use.",
    "$comment": "sourcePublic is false today. Do not describe privacy as presently auditable while this is false."
  },
  "maintainer": {
    "name": "LlamaBox AI",
    "organization": "Mythos Labs",
    "repository": "https://github.com/llamabox-ai"
  },
  "contacts": {
    "beta": "work.aalhad@gmail.com",
    "support": "work.aalhad@gmail.com",
    "business": "work.aalhad@gmail.com",
    "privacy": "work.aalhad@gmail.com"
  },
  "verifiedMetrics": {
    "$comment": "All null until measured on real hardware. Never estimate.",
    "devicesTested": null,
    "androidVersionsTested": null,
    "modelsTested": null,
    "completedTestSessions": null,
    "confirmedOfflineSessions": null,
    "crashFreeSessionPercent": null,
    "userCount": null,
    "downloadCount": null,
    "rating": null,
    "reviewCount": null
  },
  "claimsNotMade": [
    "Source is currently public or presently auditable",
    "Any GGUF model will run on any device",
    "CPU execution is universally faster than GPU or NPU inference",
    "Any user, rating, review or download count",
    "Any benchmark figure not measured on real hardware",
    "Working speech-to-text input",
    "Any public release date"
  ],
  "related": {
    "facts": "https://llamabox-ai.vercel.app/facts.html",
    "status": "https://llamabox-ai.vercel.app/status.html",
    "networkBehavior": "https://llamabox-ai.vercel.app/network-behavior.html",
    "testedDevices": "https://llamabox-ai.vercel.app/tested-devices.html",
    "benchmarks": "https://llamabox-ai.vercel.app/benchmarks.json",
    "architecture": "https://llamabox-ai.vercel.app/architecture.html",
    "llmsTxt": "https://llamabox-ai.vercel.app/llms.txt"
  }
}