@agentproto/model-catalog 0.2.0

package/dist/chunk-WRUWNKDB.mjs

@@ -0,0 +1,521 @@
+/**
+ * @agentproto/model-catalog v0.1.0
+ * Unified model catalog: LLM, image, video, audio entries + cost dispatcher + picker.
+ * Zod-only OSS core; @agstudio/model-catalog layers billing/access/BYOK on top.
+ */
+
+// src/audio/catalog.ts
+var AUDIO_MODEL_CATALOG = {
+  // ─── Speech-to-speech / realtime ────────────────────────────────────────
+  // Provider runs the LLM in-session (no separate STT+TTS billing). These
+  // bill on active session duration — `per_minute` reuses the same cost
+  // path STT uses, fed `seconds` from the voice pod's session timer.
+  // baseCost = blended USD/min (audio in+out), point-in-time; refresh via
+  // the drift-check script. agentVisible:false — these are call engines,
+  // not agent-selectable tool models.
+  // `id` is our stable internal billing id (wired into the voice engine
+  // descriptors); `providerId` is the live provider model string —
+  // refresh it as providers ship new realtime models.
+  // Gemini Live — the three provider IDs the adapter accepts. This
+  // catalog is the source of truth for those ids + the default; the
+  // adapter / tts-providers / env should resolve from here, not
+  // re-hardcode. Default = flash (low latency, the call default).
+  "google/gemini-live": {
+    id: "google/gemini-live",
+    name: "Gemini Live 2.5 Flash",
+    providerId: "gemini-2.5-flash-preview-tts",
+    provider: "google",
+    capabilities: {
+      s2s: true,
+      languages: ["multi"],
+      streaming: true
+    },
+    pricing: {
+      billingUnit: "per_minute",
+      // ~$0.10/min blended audio in+out (Gemini Live token rates).
+      baseCost: 0.1
+    },
+    description: "Google Gemini Live realtime s2s \u2014 low latency (default)",
+    agentVisible: false
+  },
+  "google/gemini-live-pro": {
+    id: "google/gemini-live-pro",
+    name: "Gemini Live 2.5 Pro",
+    providerId: "gemini-2.5-pro-preview-tts",
+    provider: "google",
+    capabilities: {
+      s2s: true,
+      languages: ["multi"],
+      streaming: true
+    },
+    pricing: {
+      billingUnit: "per_minute",
+      // Heavier/higher-quality tier — ~2× flash.
+      baseCost: 0.2
+    },
+    description: "Google Gemini Live realtime s2s \u2014 higher quality",
+    agentVisible: false
+  },
+  "google/gemini-live-3.1-flash": {
+    id: "google/gemini-live-3.1-flash",
+    name: "Gemini Live 3.1 Flash",
+    providerId: "gemini-3.1-flash-tts-preview",
+    provider: "google",
+    capabilities: {
+      s2s: true,
+      languages: ["multi"],
+      streaming: true
+    },
+    pricing: {
+      billingUnit: "per_minute",
+      baseCost: 0.1
+    },
+    description: "Google Gemini Live realtime s2s \u2014 3.1 flash preview",
+    agentVisible: false
+  },
+  "openai/gpt-realtime": {
+    id: "openai/gpt-realtime",
+    name: "OpenAI Realtime (gpt-realtime-1.5)",
+    // gpt-realtime-1.5 = best audio-in/audio-out realtime model.
+    // Variants: gpt-realtime-2 (reasoning), gpt-realtime-mini (cheap),
+    // gpt-realtime-translate (s2s translation).
+    providerId: "gpt-realtime-1.5",
+    provider: "openai",
+    capabilities: {
+      s2s: true,
+      languages: ["multi"],
+      streaming: true
+    },
+    pricing: {
+      billingUnit: "per_minute",
+      // ~$0.30/min blended (audio in + out). Verify via drift-check.
+      baseCost: 0.3
+    },
+    description: "OpenAI realtime speech-to-speech (best audio in/out)",
+    agentVisible: false
+  },
+  "openai/gpt-realtime-mini": {
+    id: "openai/gpt-realtime-mini",
+    name: "OpenAI Realtime mini",
+    providerId: "gpt-realtime-mini",
+    provider: "openai",
+    capabilities: {
+      s2s: true,
+      languages: ["multi"],
+      streaming: true
+    },
+    pricing: {
+      billingUnit: "per_minute",
+      // Cost-efficient realtime tier — ~⅓ of gpt-realtime-1.5.
+      baseCost: 0.1
+    },
+    description: "Cost-efficient OpenAI realtime speech-to-speech",
+    agentVisible: false
+  },
+  "elevenlabs/conversational": {
+    id: "elevenlabs/conversational",
+    name: "ElevenLabs Conversational",
+    providerId: "conversational-v1",
+    provider: "elevenlabs",
+    capabilities: {
+      s2s: true,
+      languages: ["multi"],
+      streaming: true
+    },
+    pricing: {
+      billingUnit: "per_minute",
+      // ElevenLabs Conversational AI ~$0.10/min on the business tier.
+      baseCost: 0.1
+    },
+    description: "ElevenLabs Conversational AI (agent-id s2s flow)",
+    agentVisible: false
+  },
+  // ─── ElevenLabs TTS ─────────────────────────────────────────────────────
+  // Source: https://elevenlabs.io/pricing (per 1k chars rates derived from
+  // tiered credit packages — verified against direct ElevenLabs API quotes).
+  "elevenlabs/multilingual-v2": {
+    id: "elevenlabs/multilingual-v2",
+    name: "ElevenLabs Multilingual v2",
+    providerId: "eleven_multilingual_v2",
+    provider: "elevenlabs",
+    capabilities: {
+      tts: true,
+      languages: ["multi"],
+      voiceCloning: true,
+      streaming: true
+    },
+    pricing: {
+      billingUnit: "per_1k_chars",
+      baseCost: 0.3,
+      creditCost: 30
+    },
+    description: "Premium quality, 29 languages, voice cloning",
+    agentVisible: true
+  },
+  "elevenlabs/turbo-v2.5": {
+    id: "elevenlabs/turbo-v2.5",
+    name: "ElevenLabs Turbo v2.5",
+    providerId: "eleven_turbo_v2_5",
+    provider: "elevenlabs",
+    capabilities: {
+      tts: true,
+      languages: ["multi"],
+      voiceCloning: true,
+      streaming: true
+    },
+    pricing: {
+      billingUnit: "per_1k_chars",
+      baseCost: 0.15,
+      creditCost: 15
+    },
+    description: "Mid-tier \u2014 half the cost of Multilingual, ~300ms latency",
+    agentVisible: true
+  },
+  "elevenlabs/flash-v2.5": {
+    id: "elevenlabs/flash-v2.5",
+    name: "ElevenLabs Flash v2.5",
+    providerId: "eleven_flash_v2_5",
+    provider: "elevenlabs",
+    capabilities: {
+      tts: true,
+      languages: ["multi"],
+      voiceCloning: true,
+      streaming: true
+    },
+    pricing: {
+      billingUnit: "per_1k_chars",
+      baseCost: 0.075,
+      creditCost: 8
+    },
+    description: "Lowest cost, ~75ms latency \u2014 voice-chat tier",
+    agentVisible: true
+  },
+  // ─── ElevenLabs STT ─────────────────────────────────────────────────────
+  "elevenlabs/scribe-v1": {
+    id: "elevenlabs/scribe-v1",
+    name: "ElevenLabs Scribe v1",
+    providerId: "scribe_v1",
+    provider: "elevenlabs",
+    capabilities: {
+      stt: true,
+      languages: ["multi"],
+      timestamps: true,
+      diarization: true,
+      languageDetection: true
+    },
+    pricing: {
+      billingUnit: "per_minute",
+      baseCost: 667e-5,
+      creditCost: 1
+    },
+    description: "ElevenLabs transcription with diarization",
+    agentVisible: true
+  },
+  // ─── OpenAI TTS ─────────────────────────────────────────────────────────
+  // Source: https://openai.com/api/pricing/ — $15/1M chars (tts-1), $30 (hd).
+  "openai/tts-1": {
+    id: "openai/tts-1",
+    name: "OpenAI TTS-1",
+    providerId: "tts-1",
+    provider: "openai",
+    capabilities: {
+      tts: true,
+      languages: ["multi"],
+      streaming: true
+    },
+    pricing: {
+      billingUnit: "per_1k_chars",
+      baseCost: 0.015,
+      creditCost: 2
+    },
+    description: "OpenAI text-to-speech (standard)",
+    agentVisible: true
+  },
+  "openai/tts-1-hd": {
+    id: "openai/tts-1-hd",
+    name: "OpenAI TTS-1 HD",
+    providerId: "tts-1-hd",
+    provider: "openai",
+    capabilities: {
+      tts: true,
+      languages: ["multi"],
+      streaming: true
+    },
+    pricing: {
+      billingUnit: "per_1k_chars",
+      baseCost: 0.03,
+      creditCost: 4
+    },
+    description: "OpenAI text-to-speech (HD)",
+    agentVisible: true
+  },
+  "openai/gpt-4o-mini-tts": {
+    id: "openai/gpt-4o-mini-tts",
+    name: "OpenAI GPT-4o mini TTS",
+    providerId: "gpt-4o-mini-tts",
+    provider: "openai",
+    capabilities: {
+      tts: true,
+      languages: ["multi"],
+      streaming: true
+    },
+    pricing: {
+      billingUnit: "per_1k_chars",
+      // GPT-4o mini TTS — steerable speech, billed per char.
+      baseCost: 0.015
+    },
+    description: "OpenAI GPT-4o mini text-to-speech (steerable)",
+    agentVisible: true
+  },
+  // ─── OpenAI STT ─────────────────────────────────────────────────────────
+  "openai/whisper-1": {
+    id: "openai/whisper-1",
+    name: "OpenAI Whisper",
+    providerId: "whisper-1",
+    provider: "openai",
+    capabilities: {
+      stt: true,
+      languages: ["multi"],
+      timestamps: true,
+      languageDetection: true
+    },
+    pricing: {
+      billingUnit: "per_minute",
+      baseCost: 6e-3,
+      creditCost: 1
+    },
+    description: "OpenAI Whisper transcription",
+    agentVisible: true
+  },
+  // Source: https://openai.com/api/pricing/ (verified via costgoat.com/pricing/
+  // openai-transcription, fetched 2026-05-29) — gpt-4o-transcribe ~$0.006/min,
+  // gpt-4o-mini-transcribe ~$0.003/min (token-based billing exposed per-minute).
+  "openai/gpt-4o-transcribe": {
+    id: "openai/gpt-4o-transcribe",
+    name: "OpenAI GPT-4o Transcribe",
+    providerId: "gpt-4o-transcribe",
+    provider: "openai",
+    capabilities: {
+      stt: true,
+      languages: ["multi"],
+      timestamps: true,
+      languageDetection: true
+    },
+    pricing: {
+      billingUnit: "per_minute",
+      baseCost: 6e-3
+    },
+    description: "OpenAI GPT-4o transcription \u2014 higher accuracy than Whisper",
+    agentVisible: true
+  },
+  "openai/gpt-4o-mini-transcribe": {
+    id: "openai/gpt-4o-mini-transcribe",
+    name: "OpenAI GPT-4o mini Transcribe",
+    providerId: "gpt-4o-mini-transcribe",
+    provider: "openai",
+    capabilities: {
+      stt: true,
+      languages: ["multi"],
+      timestamps: true,
+      languageDetection: true
+    },
+    pricing: {
+      billingUnit: "per_minute",
+      baseCost: 3e-3
+    },
+    description: "OpenAI GPT-4o mini transcription \u2014 cheapest OpenAI STT",
+    agentVisible: true
+  },
+  // ─── MiniMax TTS ────────────────────────────────────────────────────────
+  // Voices live in `@agentproto/model-catalog/voice`; pricing keyed on model.
+  "minimax/speech-02-hd": {
+    id: "minimax/speech-02-hd",
+    name: "MiniMax Speech 02 HD",
+    providerId: "speech-02-hd",
+    provider: "minimax",
+    capabilities: {
+      tts: true,
+      languages: ["multi"],
+      voiceCloning: true,
+      streaming: true
+    },
+    pricing: {
+      billingUnit: "per_1k_chars",
+      baseCost: 0.018,
+      creditCost: 2
+    },
+    description: "MiniMax HD voice synthesis with cloning",
+    agentVisible: true
+  },
+  "minimax/speech-02-turbo": {
+    id: "minimax/speech-02-turbo",
+    name: "MiniMax Speech 02 Turbo",
+    providerId: "speech-02-turbo",
+    provider: "minimax",
+    capabilities: {
+      tts: true,
+      languages: ["multi"],
+      voiceCloning: true,
+      streaming: true
+    },
+    pricing: {
+      billingUnit: "per_1k_chars",
+      baseCost: 0.01,
+      creditCost: 1
+    },
+    description: "MiniMax low-latency TTS",
+    agentVisible: true
+  },
+  // ─── Deepgram STT ───────────────────────────────────────────────────────
+  // Source: https://deepgram.com/pricing — Nova-3 streaming $0.0058/min,
+  // batch $0.0048/min; Nova-2 streaming $0.0043, batch $0.0036.
+  "deepgram/nova-3": {
+    id: "deepgram/nova-3",
+    name: "Deepgram Nova-3",
+    providerId: "nova-3",
+    provider: "deepgram",
+    capabilities: {
+      stt: true,
+      languages: ["multi"],
+      timestamps: true,
+      diarization: true,
+      languageDetection: true
+    },
+    pricing: {
+      billingUnit: "per_minute",
+      baseCost: 48e-4,
+      creditCost: 1
+    },
+    description: "Latest Deepgram model \u2014 best accuracy",
+    agentVisible: true
+  },
+  "deepgram/nova-2": {
+    id: "deepgram/nova-2",
+    name: "Deepgram Nova-2",
+    providerId: "nova-2",
+    provider: "deepgram",
+    capabilities: {
+      stt: true,
+      languages: ["multi"],
+      timestamps: true,
+      diarization: true,
+      languageDetection: true
+    },
+    pricing: {
+      billingUnit: "per_minute",
+      baseCost: 36e-4,
+      creditCost: 1
+    },
+    description: "Production Deepgram model \u2014 battle-tested",
+    agentVisible: true
+  },
+  // ─── AssemblyAI STT ─────────────────────────────────────────────────────
+  // Source: https://www.assemblyai.com/pricing — Universal-2 $0.37/hour,
+  // Nano $0.12/hour. Per-minute math: /60.
+  "assemblyai/universal-2": {
+    id: "assemblyai/universal-2",
+    name: "AssemblyAI Universal-2",
+    providerId: "best",
+    provider: "assemblyai",
+    capabilities: {
+      stt: true,
+      languages: ["multi"],
+      timestamps: true,
+      diarization: true,
+      languageDetection: true
+    },
+    pricing: {
+      billingUnit: "per_minute",
+      baseCost: 617e-5,
+      creditCost: 1
+    },
+    description: "AssemblyAI flagship \u2014 high accuracy, slow",
+    agentVisible: true
+  },
+  "assemblyai/nano": {
+    id: "assemblyai/nano",
+    name: "AssemblyAI Nano",
+    providerId: "nano",
+    provider: "assemblyai",
+    capabilities: {
+      stt: true,
+      languages: ["multi"],
+      timestamps: true
+    },
+    pricing: {
+      billingUnit: "per_minute",
+      baseCost: 2e-3,
+      creditCost: 1
+    },
+    description: "AssemblyAI fast tier",
+    agentVisible: true
+  },
+  // ─── Replicate-hosted STT (cheap, slow) ─────────────────────────────────
+  // Source: https://replicate.com/<owner>/<model> per-run pricing pages.
+  "replicate/incredibly-fast-whisper": {
+    id: "replicate/incredibly-fast-whisper",
+    name: "Incredibly Fast Whisper (Replicate)",
+    providerId: "vaibhavs10/incredibly-fast-whisper",
+    provider: "replicate",
+    capabilities: {
+      stt: true,
+      languages: ["multi"],
+      timestamps: true,
+      languageDetection: true
+    },
+    pricing: {
+      billingUnit: "per_minute",
+      baseCost: 23e-4,
+      creditCost: 1
+    },
+    description: "Whisper-large-v3 on Replicate, fast-decoded",
+    agentVisible: true
+  },
+  // ─── Replicate-hosted TTS ───────────────────────────────────────────────
+  // Estimates per ~30s clip → per-1k-chars normalization (~150 chars / 30s).
+  "replicate/chatterbox": {
+    id: "replicate/chatterbox",
+    name: "Chatterbox (Replicate)",
+    providerId: "resemble-ai/chatterbox",
+    provider: "replicate",
+    capabilities: {
+      tts: true,
+      languages: ["en"],
+      voiceCloning: true
+    },
+    pricing: {
+      billingUnit: "per_1k_chars",
+      baseCost: 0.02,
+      creditCost: 2
+    },
+    description: "Replicate-hosted voice cloning TTS",
+    agentVisible: false
+  }
+};
+var AUDIO_MODEL_IDS = Object.keys(AUDIO_MODEL_CATALOG);
+var AGENT_AUDIO_MODEL_IDS = Object.entries(AUDIO_MODEL_CATALOG).filter(([, m]) => m.agentVisible).map(([id]) => id);
+var AGENT_TTS_MODEL_IDS = Object.entries(AUDIO_MODEL_CATALOG).filter(([, m]) => m.agentVisible && m.capabilities.tts).map(([id]) => id);
+var AGENT_STT_MODEL_IDS = Object.entries(AUDIO_MODEL_CATALOG).filter(([, m]) => m.agentVisible && m.capabilities.stt).map(([id]) => id);
+var S2S_MODEL_IDS = Object.entries(AUDIO_MODEL_CATALOG).filter(([, m]) => m.capabilities.s2s).map(([id]) => id);
+var GEMINI_LIVE_MODEL_IDS = Object.values(
+  AUDIO_MODEL_CATALOG
+).filter((m) => m.capabilities.s2s && m.provider === "google").map((m) => m.providerId);
+var DEFAULT_GEMINI_LIVE_MODEL = AUDIO_MODEL_CATALOG["google/gemini-live"]?.providerId ?? "gemini-2.5-flash-preview-tts";
+function isAcceptedGeminiLiveModel(id) {
+  return GEMINI_LIVE_MODEL_IDS.includes(id);
+}
+function generateAudioModelTable(modality) {
+  return Object.values(AUDIO_MODEL_CATALOG).filter((m) => m.agentVisible && (!modality || m.capabilities[modality])).map((m) => {
+    const unit = m.pricing.billingUnit === "per_1k_chars" ? "1k chars" : m.pricing.billingUnit === "per_minute" ? "min" : "sec";
+    const modalities = [
+      m.capabilities.tts ? "TTS" : null,
+      m.capabilities.stt ? "STT" : null,
+      m.capabilities.s2s ? "S2S" : null
+    ].filter(Boolean).join("+");
+    return `| \`${m.id}\` | ${modalities} | $${m.pricing.baseCost}/${unit} | ${m.description} |`;
+  }).join("\n");
+}
+
+export { AGENT_AUDIO_MODEL_IDS, AGENT_STT_MODEL_IDS, AGENT_TTS_MODEL_IDS, AUDIO_MODEL_CATALOG, AUDIO_MODEL_IDS, DEFAULT_GEMINI_LIVE_MODEL, GEMINI_LIVE_MODEL_IDS, S2S_MODEL_IDS, generateAudioModelTable, isAcceptedGeminiLiveModel };
+//# sourceMappingURL=chunk-WRUWNKDB.mjs.map
+//# sourceMappingURL=chunk-WRUWNKDB.mjs.map

package/dist/chunk-WRUWNKDB.mjs.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/audio/catalog.ts"],"names":[],"mappings":";;;;;;;AAiFO,IAAM,mBAAA,GAA4D;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAevE,oBAAA,EAAsB;AAAA,IACpB,EAAA,EAAI,oBAAA;AAAA,IACJ,IAAA,EAAM,uBAAA;AAAA,IACN,UAAA,EAAY,8BAAA;AAAA,IACZ,QAAA,EAAU,QAAA;AAAA,IACV,YAAA,EAAc;AAAA,MACZ,GAAA,EAAK,IAAA;AAAA,MACL,SAAA,EAAW,CAAC,OAAO,CAAA;AAAA,MACnB,SAAA,EAAW;AAAA,KACb;AAAA,IACA,OAAA,EAAS;AAAA,MACP,WAAA,EAAa,YAAA;AAAA;AAAA,MAEb,QAAA,EAAU;AAAA,KACZ;AAAA,IACA,WAAA,EAAa,8DAAA;AAAA,IACb,YAAA,EAAc;AAAA,GAChB;AAAA,EACA,wBAAA,EAA0B;AAAA,IACxB,EAAA,EAAI,wBAAA;AAAA,IACJ,IAAA,EAAM,qBAAA;AAAA,IACN,UAAA,EAAY,4BAAA;AAAA,IACZ,QAAA,EAAU,QAAA;AAAA,IACV,YAAA,EAAc;AAAA,MACZ,GAAA,EAAK,IAAA;AAAA,MACL,SAAA,EAAW,CAAC,OAAO,CAAA;AAAA,MACnB,SAAA,EAAW;AAAA,KACb;AAAA,IACA,OAAA,EAAS;AAAA,MACP,WAAA,EAAa,YAAA;AAAA;AAAA,MAEb,QAAA,EAAU;AAAA,KACZ;AAAA,IACA,WAAA,EAAa,uDAAA;AAAA,IACb,YAAA,EAAc;AAAA,GAChB;AAAA,EACA,8BAAA,EAAgC;AAAA,IAC9B,EAAA,EAAI,8BAAA;AAAA,IACJ,IAAA,EAAM,uBAAA;AAAA,IACN,UAAA,EAAY,8BAAA;AAAA,IACZ,QAAA,EAAU,QAAA;AAAA,IACV,YAAA,EAAc;AAAA,MACZ,GAAA,EAAK,IAAA;AAAA,MACL,SAAA,EAAW,CAAC,OAAO,CAAA;AAAA,MACnB,SAAA,EAAW;AAAA,KACb;AAAA,IACA,OAAA,EAAS;AAAA,MACP,WAAA,EAAa,YAAA;AAAA,MACb,QAAA,EAAU;AAAA,KACZ;AAAA,IACA,WAAA,EAAa,0DAAA;AAAA,IACb,YAAA,EAAc;AAAA,GAChB;AAAA,EACA,qBAAA,EAAuB;AAAA,IACrB,EAAA,EAAI,qBAAA;AAAA,IACJ,IAAA,EAAM,oCAAA;AAAA;AAAA;AAAA;AAAA,IAIN,UAAA,EAAY,kBAAA;AAAA,IACZ,QAAA,EAAU,QAAA;AAAA,IACV,YAAA,EAAc;AAAA,MACZ,GAAA,EAAK,IAAA;AAAA,MACL,SAAA,EAAW,CAAC,OAAO,CAAA;AAAA,MACnB,SAAA,EAAW;AAAA,KACb;AAAA,IACA,OAAA,EAAS;AAAA,MACP,WAAA,EAAa,YAAA;AAAA;AAAA,MAEb,QAAA,EAAU;AAAA,KACZ;AAAA,IACA,WAAA,EAAa,sDAAA;AAAA,IACb,YAAA,EAAc;AAAA,GAChB;AAAA,EACA,0BAAA,EAA4B;AAAA,IAC1B,EAAA,EAAI,0BAAA;AAAA,IACJ,IAAA,EAAM,sBAAA;AAAA,IACN,UAAA,EAAY,mBAAA;AAAA,IACZ,QAAA,EAAU,QAAA;AAAA,IACV,YAAA,EAAc;AAAA,MACZ,GAAA,EAAK,IAAA;AAAA,MACL,SAAA,EAAW,CAAC,OAAO,CAAA;AAAA,MACnB,SAAA,EAAW;AAAA,KACb;AAAA,IACA,OAAA,EAAS;AAAA,MACP,WAAA,EAAa,YAAA;AAAA;AAAA,MAEb,QAAA,EAAU;AAAA,KACZ;AAAA,IACA,WAAA,EAAa,iDAAA;AAAA,IACb,YAAA,EAAc;AAAA,GAChB;AAAA,EACA,2BAAA,EAA6B;AAAA,IAC3B,EAAA,EAAI,2BAAA;AAAA,IACJ,IAAA,EAAM,2BAAA;AAAA,IACN,UAAA,EAAY,mBAAA;AAAA,IACZ,QAAA,EAAU,YAAA;AAAA,IACV,YAAA,EAAc;AAAA,MACZ,GAAA,EAAK,IAAA;AAAA,MACL,SAAA,EAAW,CAAC,OAAO,CAAA;AAAA,MACnB,SAAA,EAAW;AAAA,KACb;AAAA,IACA,OAAA,EAAS;AAAA,MACP,WAAA,EAAa,YAAA;AAAA;AAAA,MAEb,QAAA,EAAU;AAAA,KACZ;AAAA,IACA,WAAA,EAAa,kDAAA;AAAA,IACb,YAAA,EAAc;AAAA,GAChB;AAAA;AAAA;AAAA;AAAA,EAKA,4BAAA,EAA8B;AAAA,IAC5B,EAAA,EAAI,4BAAA;AAAA,IACJ,IAAA,EAAM,4BAAA;AAAA,IACN,UAAA,EAAY,wBAAA;AAAA,IACZ,QAAA,EAAU,YAAA;AAAA,IACV,YAAA,EAAc;AAAA,MACZ,GAAA,EAAK,IAAA;AAAA,MACL,SAAA,EAAW,CAAC,OAAO,CAAA;AAAA,MACnB,YAAA,EAAc,IAAA;AAAA,MACd,SAAA,EAAW;AAAA,KACb;AAAA,IACA,OAAA,EAAS;AAAA,MACP,WAAA,EAAa,cAAA;AAAA,MACb,QAAA,EAAU,GAAA;AAAA,MACV,UAAA,EAAY;AAAA,KACd;AAAA,IACA,WAAA,EAAa,8CAAA;AAAA,IACb,YAAA,EAAc;AAAA,GAChB;AAAA,EACA,uBAAA,EAAyB;AAAA,IACvB,EAAA,EAAI,uBAAA;AAAA,IACJ,IAAA,EAAM,uBAAA;AAAA,IACN,UAAA,EAAY,mBAAA;AAAA,IACZ,QAAA,EAAU,YAAA;AAAA,IACV,YAAA,EAAc;AAAA,MACZ,GAAA,EAAK,IAAA;AAAA,MACL,SAAA,EAAW,CAAC,OAAO,CAAA;AAAA,MACnB,YAAA,EAAc,IAAA;AAAA,MACd,SAAA,EAAW;AAAA,KACb;AAAA,IACA,OAAA,EAAS;AAAA,MACP,WAAA,EAAa,cAAA;AAAA,MACb,QAAA,EAAU,IAAA;AAAA,MACV,UAAA,EAAY;AAAA,KACd;AAAA,IACA,WAAA,EAAa,+DAAA;AAAA,IACb,YAAA,EAAc;AAAA,GAChB;AAAA,EACA,uBAAA,EAAyB;AAAA,IACvB,EAAA,EAAI,uBAAA;AAAA,IACJ,IAAA,EAAM,uBAAA;AAAA,IACN,UAAA,EAAY,mBAAA;AAAA,IACZ,QAAA,EAAU,YAAA;AAAA,IACV,YAAA,EAAc;AAAA,MACZ,GAAA,EAAK,IAAA;AAAA,MACL,SAAA,EAAW,CAAC,OAAO,CAAA;AAAA,MACnB,YAAA,EAAc,IAAA;AAAA,MACd,SAAA,EAAW;AAAA,KACb;AAAA,IACA,OAAA,EAAS;AAAA,MACP,WAAA,EAAa,cAAA;AAAA,MACb,QAAA,EAAU,KAAA;AAAA,MACV,UAAA,EAAY;AAAA,KACd;AAAA,IACA,WAAA,EAAa,mDAAA;AAAA,IACb,YAAA,EAAc;AAAA,GAChB;AAAA;AAAA,EAGA,sBAAA,EAAwB;AAAA,IACtB,EAAA,EAAI,sBAAA;AAAA,IACJ,IAAA,EAAM,sBAAA;AAAA,IACN,UAAA,EAAY,WAAA;AAAA,IACZ,QAAA,EAAU,YAAA;AAAA,IACV,YAAA,EAAc;AAAA,MACZ,GAAA,EAAK,IAAA;AAAA,MACL,SAAA,EAAW,CAAC,OAAO,CAAA;AAAA,MACnB,UAAA,EAAY,IAAA;AAAA,MACZ,WAAA,EAAa,IAAA;AAAA,MACb,iBAAA,EAAmB;AAAA,KACrB;AAAA,IACA,OAAA,EAAS;AAAA,MACP,WAAA,EAAa,YAAA;AAAA,MACb,QAAA,EAAU,MAAA;AAAA,MACV,UAAA,EAAY;AAAA,KACd;AAAA,IACA,WAAA,EAAa,2CAAA;AAAA,IACb,YAAA,EAAc;AAAA,GAChB;AAAA;AAAA;AAAA,EAIA,cAAA,EAAgB;AAAA,IACd,EAAA,EAAI,cAAA;AAAA,IACJ,IAAA,EAAM,cAAA;AAAA,IACN,UAAA,EAAY,OAAA;AAAA,IACZ,QAAA,EAAU,QAAA;AAAA,IACV,YAAA,EAAc;AAAA,MACZ,GAAA,EAAK,IAAA;AAAA,MACL,SAAA,EAAW,CAAC,OAAO,CAAA;AAAA,MACnB,SAAA,EAAW;AAAA,KACb;AAAA,IACA,OAAA,EAAS;AAAA,MACP,WAAA,EAAa,cAAA;AAAA,MACb,QAAA,EAAU,KAAA;AAAA,MACV,UAAA,EAAY;AAAA,KACd;AAAA,IACA,WAAA,EAAa,kCAAA;AAAA,IACb,YAAA,EAAc;AAAA,GAChB;AAAA,EACA,iBAAA,EAAmB;AAAA,IACjB,EAAA,EAAI,iBAAA;AAAA,IACJ,IAAA,EAAM,iBAAA;AAAA,IACN,UAAA,EAAY,UAAA;AAAA,IACZ,QAAA,EAAU,QAAA;AAAA,IACV,YAAA,EAAc;AAAA,MACZ,GAAA,EAAK,IAAA;AAAA,MACL,SAAA,EAAW,CAAC,OAAO,CAAA;AAAA,MACnB,SAAA,EAAW;AAAA,KACb;AAAA,IACA,OAAA,EAAS;AAAA,MACP,WAAA,EAAa,cAAA;AAAA,MACb,QAAA,EAAU,IAAA;AAAA,MACV,UAAA,EAAY;AAAA,KACd;AAAA,IACA,WAAA,EAAa,4BAAA;AAAA,IACb,YAAA,EAAc;AAAA,GAChB;AAAA,EACA,wBAAA,EAA0B;AAAA,IACxB,EAAA,EAAI,wBAAA;AAAA,IACJ,IAAA,EAAM,wBAAA;AAAA,IACN,UAAA,EAAY,iBAAA;AAAA,IACZ,QAAA,EAAU,QAAA;AAAA,IACV,YAAA,EAAc;AAAA,MACZ,GAAA,EAAK,IAAA;AAAA,MACL,SAAA,EAAW,CAAC,OAAO,CAAA;AAAA,MACnB,SAAA,EAAW;AAAA,KACb;AAAA,IACA,OAAA,EAAS;AAAA,MACP,WAAA,EAAa,cAAA;AAAA;AAAA,MAEb,QAAA,EAAU;AAAA,KACZ;AAAA,IACA,WAAA,EAAa,+CAAA;AAAA,IACb,YAAA,EAAc;AAAA,GAChB;AAAA;AAAA,EAGA,kBAAA,EAAoB;AAAA,IAClB,EAAA,EAAI,kBAAA;AAAA,IACJ,IAAA,EAAM,gBAAA;AAAA,IACN,UAAA,EAAY,WAAA;AAAA,IACZ,QAAA,EAAU,QAAA;AAAA,IACV,YAAA,EAAc;AAAA,MACZ,GAAA,EAAK,IAAA;AAAA,MACL,SAAA,EAAW,CAAC,OAAO,CAAA;AAAA,MACnB,UAAA,EAAY,IAAA;AAAA,MACZ,iBAAA,EAAmB;AAAA,KACrB;AAAA,IACA,OAAA,EAAS;AAAA,MACP,WAAA,EAAa,YAAA;AAAA,MACb,QAAA,EAAU,IAAA;AAAA,MACV,UAAA,EAAY;AAAA,KACd;AAAA,IACA,WAAA,EAAa,8BAAA;AAAA,IACb,YAAA,EAAc;AAAA,GAChB;AAAA;AAAA;AAAA;AAAA,EAIA,0BAAA,EAA4B;AAAA,IAC1B,EAAA,EAAI,0BAAA;AAAA,IACJ,IAAA,EAAM,0BAAA;AAAA,IACN,UAAA,EAAY,mBAAA;AAAA,IACZ,QAAA,EAAU,QAAA;AAAA,IACV,YAAA,EAAc;AAAA,MACZ,GAAA,EAAK,IAAA;AAAA,MACL,SAAA,EAAW,CAAC,OAAO,CAAA;AAAA,MACnB,UAAA,EAAY,IAAA;AAAA,MACZ,iBAAA,EAAmB;AAAA,KACrB;AAAA,IACA,OAAA,EAAS;AAAA,MACP,WAAA,EAAa,YAAA;AAAA,MACb,QAAA,EAAU;AAAA,KACZ;AAAA,IACA,WAAA,EAAa,iEAAA;AAAA,IACb,YAAA,EAAc;AAAA,GAChB;AAAA,EACA,+BAAA,EAAiC;AAAA,IAC/B,EAAA,EAAI,+BAAA;AAAA,IACJ,IAAA,EAAM,+BAAA;AAAA,IACN,UAAA,EAAY,wBAAA;AAAA,IACZ,QAAA,EAAU,QAAA;AAAA,IACV,YAAA,EAAc;AAAA,MACZ,GAAA,EAAK,IAAA;AAAA,MACL,SAAA,EAAW,CAAC,OAAO,CAAA;AAAA,MACnB,UAAA,EAAY,IAAA;AAAA,MACZ,iBAAA,EAAmB;AAAA,KACrB;AAAA,IACA,OAAA,EAAS;AAAA,MACP,WAAA,EAAa,YAAA;AAAA,MACb,QAAA,EAAU;AAAA,KACZ;AAAA,IACA,WAAA,EAAa,6DAAA;AAAA,IACb,YAAA,EAAc;AAAA,GAChB;AAAA;AAAA;AAAA,EAIA,sBAAA,EAAwB;AAAA,IACtB,EAAA,EAAI,sBAAA;AAAA,IACJ,IAAA,EAAM,sBAAA;AAAA,IACN,UAAA,EAAY,cAAA;AAAA,IACZ,QAAA,EAAU,SAAA;AAAA,IACV,YAAA,EAAc;AAAA,MACZ,GAAA,EAAK,IAAA;AAAA,MACL,SAAA,EAAW,CAAC,OAAO,CAAA;AAAA,MACnB,YAAA,EAAc,IAAA;AAAA,MACd,SAAA,EAAW;AAAA,KACb;AAAA,IACA,OAAA,EAAS;AAAA,MACP,WAAA,EAAa,cAAA;AAAA,MACb,QAAA,EAAU,KAAA;AAAA,MACV,UAAA,EAAY;AAAA,KACd;AAAA,IACA,WAAA,EAAa,yCAAA;AAAA,IACb,YAAA,EAAc;AAAA,GAChB;AAAA,EACA,yBAAA,EAA2B;AAAA,IACzB,EAAA,EAAI,yBAAA;AAAA,IACJ,IAAA,EAAM,yBAAA;AAAA,IACN,UAAA,EAAY,iBAAA;AAAA,IACZ,QAAA,EAAU,SAAA;AAAA,IACV,YAAA,EAAc;AAAA,MACZ,GAAA,EAAK,IAAA;AAAA,MACL,SAAA,EAAW,CAAC,OAAO,CAAA;AAAA,MACnB,YAAA,EAAc,IAAA;AAAA,MACd,SAAA,EAAW;AAAA,KACb;AAAA,IACA,OAAA,EAAS;AAAA,MACP,WAAA,EAAa,cAAA;AAAA,MACb,QAAA,EAAU,IAAA;AAAA,MACV,UAAA,EAAY;AAAA,KACd;AAAA,IACA,WAAA,EAAa,yBAAA;AAAA,IACb,YAAA,EAAc;AAAA,GAChB;AAAA;AAAA;AAAA;AAAA,EAKA,iBAAA,EAAmB;AAAA,IACjB,EAAA,EAAI,iBAAA;AAAA,IACJ,IAAA,EAAM,iBAAA;AAAA,IACN,UAAA,EAAY,QAAA;AAAA,IACZ,QAAA,EAAU,UAAA;AAAA,IACV,YAAA,EAAc;AAAA,MACZ,GAAA,EAAK,IAAA;AAAA,MACL,SAAA,EAAW,CAAC,OAAO,CAAA;AAAA,MACnB,UAAA,EAAY,IAAA;AAAA,MACZ,WAAA,EAAa,IAAA;AAAA,MACb,iBAAA,EAAmB;AAAA,KACrB;AAAA,IACA,OAAA,EAAS;AAAA,MACP,WAAA,EAAa,YAAA;AAAA,MACb,QAAA,EAAU,KAAA;AAAA,MACV,UAAA,EAAY;AAAA,KACd;AAAA,IACA,WAAA,EAAa,4CAAA;AAAA,IACb,YAAA,EAAc;AAAA,GAChB;AAAA,EACA,iBAAA,EAAmB;AAAA,IACjB,EAAA,EAAI,iBAAA;AAAA,IACJ,IAAA,EAAM,iBAAA;AAAA,IACN,UAAA,EAAY,QAAA;AAAA,IACZ,QAAA,EAAU,UAAA;AAAA,IACV,YAAA,EAAc;AAAA,MACZ,GAAA,EAAK,IAAA;AAAA,MACL,SAAA,EAAW,CAAC,OAAO,CAAA;AAAA,MACnB,UAAA,EAAY,IAAA;AAAA,MACZ,WAAA,EAAa,IAAA;AAAA,MACb,iBAAA,EAAmB;AAAA,KACrB;AAAA,IACA,OAAA,EAAS;AAAA,MACP,WAAA,EAAa,YAAA;AAAA,MACb,QAAA,EAAU,KAAA;AAAA,MACV,UAAA,EAAY;AAAA,KACd;AAAA,IACA,WAAA,EAAa,gDAAA;AAAA,IACb,YAAA,EAAc;AAAA,GAChB;AAAA;AAAA;AAAA;AAAA,EAKA,wBAAA,EAA0B;AAAA,IACxB,EAAA,EAAI,wBAAA;AAAA,IACJ,IAAA,EAAM,wBAAA;AAAA,IACN,UAAA,EAAY,MAAA;AAAA,IACZ,QAAA,EAAU,YAAA;AAAA,IACV,YAAA,EAAc;AAAA,MACZ,GAAA,EAAK,IAAA;AAAA,MACL,SAAA,EAAW,CAAC,OAAO,CAAA;AAAA,MACnB,UAAA,EAAY,IAAA;AAAA,MACZ,WAAA,EAAa,IAAA;AAAA,MACb,iBAAA,EAAmB;AAAA,KACrB;AAAA,IACA,OAAA,EAAS;AAAA,MACP,WAAA,EAAa,YAAA;AAAA,MACb,QAAA,EAAU,MAAA;AAAA,MACV,UAAA,EAAY;AAAA,KACd;AAAA,IACA,WAAA,EAAa,gDAAA;AAAA,IACb,YAAA,EAAc;AAAA,GAChB;AAAA,EACA,iBAAA,EAAmB;AAAA,IACjB,EAAA,EAAI,iBAAA;AAAA,IACJ,IAAA,EAAM,iBAAA;AAAA,IACN,UAAA,EAAY,MAAA;AAAA,IACZ,QAAA,EAAU,YAAA;AAAA,IACV,YAAA,EAAc;AAAA,MACZ,GAAA,EAAK,IAAA;AAAA,MACL,SAAA,EAAW,CAAC,OAAO,CAAA;AAAA,MACnB,UAAA,EAAY;AAAA,KACd;AAAA,IACA,OAAA,EAAS;AAAA,MACP,WAAA,EAAa,YAAA;AAAA,MACb,QAAA,EAAU,IAAA;AAAA,MACV,UAAA,EAAY;AAAA,KACd;AAAA,IACA,WAAA,EAAa,sBAAA;AAAA,IACb,YAAA,EAAc;AAAA,GAChB;AAAA;AAAA;AAAA,EAIA,mCAAA,EAAqC;AAAA,IACnC,EAAA,EAAI,mCAAA;AAAA,IACJ,IAAA,EAAM,qCAAA;AAAA,IACN,UAAA,EAAY,oCAAA;AAAA,IACZ,QAAA,EAAU,WAAA;AAAA,IACV,YAAA,EAAc;AAAA,MACZ,GAAA,EAAK,IAAA;AAAA,MACL,SAAA,EAAW,CAAC,OAAO,CAAA;AAAA,MACnB,UAAA,EAAY,IAAA;AAAA,MACZ,iBAAA,EAAmB;AAAA,KACrB;AAAA,IACA,OAAA,EAAS;AAAA,MACP,WAAA,EAAa,YAAA;AAAA,MACb,QAAA,EAAU,KAAA;AAAA,MACV,UAAA,EAAY;AAAA,KACd;AAAA,IACA,WAAA,EAAa,6CAAA;AAAA,IACb,YAAA,EAAc;AAAA,GAChB;AAAA;AAAA;AAAA,EAIA,sBAAA,EAAwB;AAAA,IACtB,EAAA,EAAI,sBAAA;AAAA,IACJ,IAAA,EAAM,wBAAA;AAAA,IACN,UAAA,EAAY,wBAAA;AAAA,IACZ,QAAA,EAAU,WAAA;AAAA,IACV,YAAA,EAAc;AAAA,MACZ,GAAA,EAAK,IAAA;AAAA,MACL,SAAA,EAAW,CAAC,IAAI,CAAA;AAAA,MAChB,YAAA,EAAc;AAAA,KAChB;AAAA,IACA,OAAA,EAAS;AAAA,MACP,WAAA,EAAa,cAAA;AAAA,MACb,QAAA,EAAU,IAAA;AAAA,MACV,UAAA,EAAY;AAAA,KACd;AAAA,IACA,WAAA,EAAa,oCAAA;AAAA,IACb,YAAA,EAAc;AAAA;AAElB;AAIO,IAAM,eAAA,GAAkB,MAAA,CAAO,IAAA,CAAK,mBAAmB;AAKvD,IAAM,wBAAwB,MAAA,CAAO,OAAA,CAAQ,mBAAmB,CAAA,CACpE,MAAA,CAAO,CAAC,GAAG,CAAC,CAAA,KAAM,CAAA,CAAE,YAAY,CAAA,CAChC,GAAA,CAAI,CAAC,CAAC,EAAE,MAAM,EAAE;AAEZ,IAAM,mBAAA,GAAsB,OAAO,OAAA,CAAQ,mBAAmB,EAClE,MAAA,CAAO,CAAC,GAAG,CAAC,CAAA,KAAM,EAAE,YAAA,IAAgB,CAAA,CAAE,aAAa,GAAG,CAAA,CACtD,IAAI,CAAC,CAAC,EAAE,CAAA,KAAM,EAAE;AAEZ,IAAM,mBAAA,GAAsB,OAAO,OAAA,CAAQ,mBAAmB,EAClE,MAAA,CAAO,CAAC,GAAG,CAAC,CAAA,KAAM,EAAE,YAAA,IAAgB,CAAA,CAAE,aAAa,GAAG,CAAA,CACtD,IAAI,CAAC,CAAC,EAAE,CAAA,KAAM,EAAE;AAQZ,IAAM,aAAA,GAAgB,OAAO,OAAA,CAAQ,mBAAmB,EAC5D,MAAA,CAAO,CAAC,GAAG,CAAC,MAAM,CAAA,CAAE,YAAA,CAAa,GAAG,CAAA,CACpC,GAAA,CAAI,CAAC,CAAC,EAAE,MAAM,EAAE;AAIZ,IAAM,wBAAkC,MAAA,CAAO,MAAA;AAAA,EACpD;AACF,CAAA,CACG,MAAA,CAAO,CAAA,CAAA,KAAK,CAAA,CAAE,YAAA,CAAa,GAAA,IAAO,CAAA,CAAE,QAAA,KAAa,QAAQ,CAAA,CACzD,GAAA,CAAI,CAAA,CAAA,KAAK,CAAA,CAAE,UAAU;AAGjB,IAAM,yBAAA,GACX,mBAAA,CAAoB,oBAAoB,CAAA,EAAG,UAAA,IAC3C;AAGK,SAAS,0BAA0B,EAAA,EAAqB;AAC7D,EAAA,OAAO,qBAAA,CAAsB,SAAS,EAAE,CAAA;AAC1C;AAGO,SAAS,wBAAwB,QAAA,EAAkC;AACxE,EAAA,OAAO,OAAO,MAAA,CAAO,mBAAmB,CAAA,CACrC,MAAA,CAAO,OAAK,CAAA,CAAE,YAAA,KAAiB,CAAC,QAAA,IAAY,EAAE,YAAA,CAAa,QAAQ,CAAA,CAAE,CAAA,CACrE,IAAI,CAAA,CAAA,KAAK;AACR,IAAA,MAAM,IAAA,GACJ,CAAA,CAAE,OAAA,CAAQ,WAAA,KAAgB,cAAA,GACtB,aACA,CAAA,CAAE,OAAA,CAAQ,WAAA,KAAgB,YAAA,GACxB,KAAA,GACA,KAAA;AACR,IAAA,MAAM,UAAA,GAAa;AAAA,MACjB,CAAA,CAAE,YAAA,CAAa,GAAA,GAAM,KAAA,GAAQ,IAAA;AAAA,MAC7B,CAAA,CAAE,YAAA,CAAa,GAAA,GAAM,KAAA,GAAQ,IAAA;AAAA,MAC7B,CAAA,CAAE,YAAA,CAAa,GAAA,GAAM,KAAA,GAAQ;AAAA,KAC/B,CACG,MAAA,CAAO,OAAO,CAAA,CACd,KAAK,GAAG,CAAA;AACX,IAAA,OAAO,CAAA,IAAA,EAAO,CAAA,CAAE,EAAE,CAAA,KAAA,EAAQ,UAAU,CAAA,IAAA,EAAO,CAAA,CAAE,OAAA,CAAQ,QAAQ,CAAA,CAAA,EAAI,IAAI,CAAA,GAAA,EAAM,EAAE,WAAW,CAAA,EAAA,CAAA;AAAA,EAC1F,CAAC,CAAA,CACA,IAAA,CAAK,IAAI,CAAA;AACd","file":"chunk-WRUWNKDB.mjs","sourcesContent":["/**\n * Audio Model Catalog — TTS + STT provider models.\n *\n * Pairs with the voice catalog (`@agentproto/model-catalog/voice`) but a\n * separate concern: voices describe *what the output sounds like*, models\n * here describe *how the provider bills*. A single TTS model (e.g.\n * ElevenLabs `multilingual-v2`) can render hundreds of voices at the\n * same rate.\n *\n * Billing follows the per-kind dispatcher in `cost/index.ts`:\n *   - TTS bills `per_1k_chars` (multiply characters/1000 × creditCost)\n *   - STT bills `per_minute` (multiply ceil(seconds/60) × creditCost)\n *\n * Adding a model = 1 entry here + (if used by an integration) a\n * provider call-site that passes the model id to the cost dispatcher\n * so `calculateCost(modelId, { kind: \"audio\", characters / seconds })`\n * returns real numbers instead of the placeholder.\n *\n * Prices are point-in-time (refresh monthly via the drift-check\n * script). Sources noted per entry — production cost ≠ retail; we\n * markup ~1.25-1.5× into `creditCost` (displayed credits, 1 cred ≈\n * €0.01) consistent with the image/video catalogs.\n */\n\nimport type { AudioCapabilities } from \"../schema/audio.js\"\n\n/**\n * Audio MODALITY labels — what an audio model does. A model may do more\n * than one (e.g. a combined engine, or a provider exposing both cascade\n * TTS and conversational s2s), so these are boolean flags on\n * `AudioCapabilities`, not a single discriminator. The model's catalog\n * `kind` is always the family `\"audio\"`; modality lives here.\n *   - `tts` — text → speech\n *   - `stt` — speech → text\n *   - `s2s` — speech ↔ speech (one model; distinct from a tts+llm+stt cascade)\n */\nexport type AudioModality = \"tts\" | \"stt\" | \"s2s\"\n\nexport type AudioBillingUnit = \"per_1k_chars\" | \"per_minute\" | \"per_second\"\n\nexport type AudioProvider =\n  | \"elevenlabs\"\n  | \"minimax\"\n  | \"openai\"\n  | \"replicate\"\n  | \"deepgram\"\n  | \"assemblyai\"\n  | \"google\"\n\nexport interface AudioPricing {\n  /** How the provider charges (drives the cost dispatcher math). */\n  billingUnit: AudioBillingUnit\n  /** Production cost in USD per `billingUnit`. */\n  baseCost: number\n  /**\n   * @deprecated Credits derived at call time via\n   * `computeCenticredits({ baseCostUsd: baseCost, category: \"audio\" })`.\n   * Kept optional for back-compat with existing entries.\n   */\n  creditCost?: number\n  /** Bypass markup formula and charge exactly this many credits per billingUnit. */\n  overrideCreditCost?: number\n}\n\n// `AudioCapabilities` is owned by the zod schema (`../schema/audio.ts`) —\n// single source of truth (imported at the top). Re-exported so existing\n// `@agentproto/model-catalog/audio` consumers keep importing it from here.\nexport type { AudioCapabilities }\n\nexport interface AudioModelDefinition {\n  id: string\n  name: string\n  /** Provider-facing identifier (e.g. `eleven_multilingual_v2`, `nova-3`). */\n  providerId: string\n  provider: AudioProvider\n  capabilities: AudioCapabilities\n  pricing: AudioPricing\n  description: string\n  agentVisible: boolean\n}\n\nexport const AUDIO_MODEL_CATALOG: Record<string, AudioModelDefinition> = {\n  // ─── Speech-to-speech / realtime ────────────────────────────────────────\n  // Provider runs the LLM in-session (no separate STT+TTS billing). These\n  // bill on active session duration — `per_minute` reuses the same cost\n  // path STT uses, fed `seconds` from the voice pod's session timer.\n  // baseCost = blended USD/min (audio in+out), point-in-time; refresh via\n  // the drift-check script. agentVisible:false — these are call engines,\n  // not agent-selectable tool models.\n  // `id` is our stable internal billing id (wired into the voice engine\n  // descriptors); `providerId` is the live provider model string —\n  // refresh it as providers ship new realtime models.\n  // Gemini Live — the three provider IDs the adapter accepts. This\n  // catalog is the source of truth for those ids + the default; the\n  // adapter / tts-providers / env should resolve from here, not\n  // re-hardcode. Default = flash (low latency, the call default).\n  \"google/gemini-live\": {\n    id: \"google/gemini-live\",\n    name: \"Gemini Live 2.5 Flash\",\n    providerId: \"gemini-2.5-flash-preview-tts\",\n    provider: \"google\",\n    capabilities: {\n      s2s: true,\n      languages: [\"multi\"],\n      streaming: true,\n    },\n    pricing: {\n      billingUnit: \"per_minute\",\n      // ~$0.10/min blended audio in+out (Gemini Live token rates).\n      baseCost: 0.1,\n    },\n    description: \"Google Gemini Live realtime s2s — low latency (default)\",\n    agentVisible: false,\n  },\n  \"google/gemini-live-pro\": {\n    id: \"google/gemini-live-pro\",\n    name: \"Gemini Live 2.5 Pro\",\n    providerId: \"gemini-2.5-pro-preview-tts\",\n    provider: \"google\",\n    capabilities: {\n      s2s: true,\n      languages: [\"multi\"],\n      streaming: true,\n    },\n    pricing: {\n      billingUnit: \"per_minute\",\n      // Heavier/higher-quality tier — ~2× flash.\n      baseCost: 0.2,\n    },\n    description: \"Google Gemini Live realtime s2s — higher quality\",\n    agentVisible: false,\n  },\n  \"google/gemini-live-3.1-flash\": {\n    id: \"google/gemini-live-3.1-flash\",\n    name: \"Gemini Live 3.1 Flash\",\n    providerId: \"gemini-3.1-flash-tts-preview\",\n    provider: \"google\",\n    capabilities: {\n      s2s: true,\n      languages: [\"multi\"],\n      streaming: true,\n    },\n    pricing: {\n      billingUnit: \"per_minute\",\n      baseCost: 0.1,\n    },\n    description: \"Google Gemini Live realtime s2s — 3.1 flash preview\",\n    agentVisible: false,\n  },\n  \"openai/gpt-realtime\": {\n    id: \"openai/gpt-realtime\",\n    name: \"OpenAI Realtime (gpt-realtime-1.5)\",\n    // gpt-realtime-1.5 = best audio-in/audio-out realtime model.\n    // Variants: gpt-realtime-2 (reasoning), gpt-realtime-mini (cheap),\n    // gpt-realtime-translate (s2s translation).\n    providerId: \"gpt-realtime-1.5\",\n    provider: \"openai\",\n    capabilities: {\n      s2s: true,\n      languages: [\"multi\"],\n      streaming: true,\n    },\n    pricing: {\n      billingUnit: \"per_minute\",\n      // ~$0.30/min blended (audio in + out). Verify via drift-check.\n      baseCost: 0.3,\n    },\n    description: \"OpenAI realtime speech-to-speech (best audio in/out)\",\n    agentVisible: false,\n  },\n  \"openai/gpt-realtime-mini\": {\n    id: \"openai/gpt-realtime-mini\",\n    name: \"OpenAI Realtime mini\",\n    providerId: \"gpt-realtime-mini\",\n    provider: \"openai\",\n    capabilities: {\n      s2s: true,\n      languages: [\"multi\"],\n      streaming: true,\n    },\n    pricing: {\n      billingUnit: \"per_minute\",\n      // Cost-efficient realtime tier — ~⅓ of gpt-realtime-1.5.\n      baseCost: 0.1,\n    },\n    description: \"Cost-efficient OpenAI realtime speech-to-speech\",\n    agentVisible: false,\n  },\n  \"elevenlabs/conversational\": {\n    id: \"elevenlabs/conversational\",\n    name: \"ElevenLabs Conversational\",\n    providerId: \"conversational-v1\",\n    provider: \"elevenlabs\",\n    capabilities: {\n      s2s: true,\n      languages: [\"multi\"],\n      streaming: true,\n    },\n    pricing: {\n      billingUnit: \"per_minute\",\n      // ElevenLabs Conversational AI ~$0.10/min on the business tier.\n      baseCost: 0.1,\n    },\n    description: \"ElevenLabs Conversational AI (agent-id s2s flow)\",\n    agentVisible: false,\n  },\n\n  // ─── ElevenLabs TTS ─────────────────────────────────────────────────────\n  // Source: https://elevenlabs.io/pricing (per 1k chars rates derived from\n  // tiered credit packages — verified against direct ElevenLabs API quotes).\n  \"elevenlabs/multilingual-v2\": {\n    id: \"elevenlabs/multilingual-v2\",\n    name: \"ElevenLabs Multilingual v2\",\n    providerId: \"eleven_multilingual_v2\",\n    provider: \"elevenlabs\",\n    capabilities: {\n      tts: true,\n      languages: [\"multi\"],\n      voiceCloning: true,\n      streaming: true,\n    },\n    pricing: {\n      billingUnit: \"per_1k_chars\",\n      baseCost: 0.3,\n      creditCost: 30,\n    },\n    description: \"Premium quality, 29 languages, voice cloning\",\n    agentVisible: true,\n  },\n  \"elevenlabs/turbo-v2.5\": {\n    id: \"elevenlabs/turbo-v2.5\",\n    name: \"ElevenLabs Turbo v2.5\",\n    providerId: \"eleven_turbo_v2_5\",\n    provider: \"elevenlabs\",\n    capabilities: {\n      tts: true,\n      languages: [\"multi\"],\n      voiceCloning: true,\n      streaming: true,\n    },\n    pricing: {\n      billingUnit: \"per_1k_chars\",\n      baseCost: 0.15,\n      creditCost: 15,\n    },\n    description: \"Mid-tier — half the cost of Multilingual, ~300ms latency\",\n    agentVisible: true,\n  },\n  \"elevenlabs/flash-v2.5\": {\n    id: \"elevenlabs/flash-v2.5\",\n    name: \"ElevenLabs Flash v2.5\",\n    providerId: \"eleven_flash_v2_5\",\n    provider: \"elevenlabs\",\n    capabilities: {\n      tts: true,\n      languages: [\"multi\"],\n      voiceCloning: true,\n      streaming: true,\n    },\n    pricing: {\n      billingUnit: \"per_1k_chars\",\n      baseCost: 0.075,\n      creditCost: 8,\n    },\n    description: \"Lowest cost, ~75ms latency — voice-chat tier\",\n    agentVisible: true,\n  },\n\n  // ─── ElevenLabs STT ─────────────────────────────────────────────────────\n  \"elevenlabs/scribe-v1\": {\n    id: \"elevenlabs/scribe-v1\",\n    name: \"ElevenLabs Scribe v1\",\n    providerId: \"scribe_v1\",\n    provider: \"elevenlabs\",\n    capabilities: {\n      stt: true,\n      languages: [\"multi\"],\n      timestamps: true,\n      diarization: true,\n      languageDetection: true,\n    },\n    pricing: {\n      billingUnit: \"per_minute\",\n      baseCost: 0.00667,\n      creditCost: 1,\n    },\n    description: \"ElevenLabs transcription with diarization\",\n    agentVisible: true,\n  },\n\n  // ─── OpenAI TTS ─────────────────────────────────────────────────────────\n  // Source: https://openai.com/api/pricing/ — $15/1M chars (tts-1), $30 (hd).\n  \"openai/tts-1\": {\n    id: \"openai/tts-1\",\n    name: \"OpenAI TTS-1\",\n    providerId: \"tts-1\",\n    provider: \"openai\",\n    capabilities: {\n      tts: true,\n      languages: [\"multi\"],\n      streaming: true,\n    },\n    pricing: {\n      billingUnit: \"per_1k_chars\",\n      baseCost: 0.015,\n      creditCost: 2,\n    },\n    description: \"OpenAI text-to-speech (standard)\",\n    agentVisible: true,\n  },\n  \"openai/tts-1-hd\": {\n    id: \"openai/tts-1-hd\",\n    name: \"OpenAI TTS-1 HD\",\n    providerId: \"tts-1-hd\",\n    provider: \"openai\",\n    capabilities: {\n      tts: true,\n      languages: [\"multi\"],\n      streaming: true,\n    },\n    pricing: {\n      billingUnit: \"per_1k_chars\",\n      baseCost: 0.03,\n      creditCost: 4,\n    },\n    description: \"OpenAI text-to-speech (HD)\",\n    agentVisible: true,\n  },\n  \"openai/gpt-4o-mini-tts\": {\n    id: \"openai/gpt-4o-mini-tts\",\n    name: \"OpenAI GPT-4o mini TTS\",\n    providerId: \"gpt-4o-mini-tts\",\n    provider: \"openai\",\n    capabilities: {\n      tts: true,\n      languages: [\"multi\"],\n      streaming: true,\n    },\n    pricing: {\n      billingUnit: \"per_1k_chars\",\n      // GPT-4o mini TTS — steerable speech, billed per char.\n      baseCost: 0.015,\n    },\n    description: \"OpenAI GPT-4o mini text-to-speech (steerable)\",\n    agentVisible: true,\n  },\n\n  // ─── OpenAI STT ─────────────────────────────────────────────────────────\n  \"openai/whisper-1\": {\n    id: \"openai/whisper-1\",\n    name: \"OpenAI Whisper\",\n    providerId: \"whisper-1\",\n    provider: \"openai\",\n    capabilities: {\n      stt: true,\n      languages: [\"multi\"],\n      timestamps: true,\n      languageDetection: true,\n    },\n    pricing: {\n      billingUnit: \"per_minute\",\n      baseCost: 0.006,\n      creditCost: 1,\n    },\n    description: \"OpenAI Whisper transcription\",\n    agentVisible: true,\n  },\n  // Source: https://openai.com/api/pricing/ (verified via costgoat.com/pricing/\n  // openai-transcription, fetched 2026-05-29) — gpt-4o-transcribe ~$0.006/min,\n  // gpt-4o-mini-transcribe ~$0.003/min (token-based billing exposed per-minute).\n  \"openai/gpt-4o-transcribe\": {\n    id: \"openai/gpt-4o-transcribe\",\n    name: \"OpenAI GPT-4o Transcribe\",\n    providerId: \"gpt-4o-transcribe\",\n    provider: \"openai\",\n    capabilities: {\n      stt: true,\n      languages: [\"multi\"],\n      timestamps: true,\n      languageDetection: true,\n    },\n    pricing: {\n      billingUnit: \"per_minute\",\n      baseCost: 0.006,\n    },\n    description: \"OpenAI GPT-4o transcription — higher accuracy than Whisper\",\n    agentVisible: true,\n  },\n  \"openai/gpt-4o-mini-transcribe\": {\n    id: \"openai/gpt-4o-mini-transcribe\",\n    name: \"OpenAI GPT-4o mini Transcribe\",\n    providerId: \"gpt-4o-mini-transcribe\",\n    provider: \"openai\",\n    capabilities: {\n      stt: true,\n      languages: [\"multi\"],\n      timestamps: true,\n      languageDetection: true,\n    },\n    pricing: {\n      billingUnit: \"per_minute\",\n      baseCost: 0.003,\n    },\n    description: \"OpenAI GPT-4o mini transcription — cheapest OpenAI STT\",\n    agentVisible: true,\n  },\n\n  // ─── MiniMax TTS ────────────────────────────────────────────────────────\n  // Voices live in `@agentproto/model-catalog/voice`; pricing keyed on model.\n  \"minimax/speech-02-hd\": {\n    id: \"minimax/speech-02-hd\",\n    name: \"MiniMax Speech 02 HD\",\n    providerId: \"speech-02-hd\",\n    provider: \"minimax\",\n    capabilities: {\n      tts: true,\n      languages: [\"multi\"],\n      voiceCloning: true,\n      streaming: true,\n    },\n    pricing: {\n      billingUnit: \"per_1k_chars\",\n      baseCost: 0.018,\n      creditCost: 2,\n    },\n    description: \"MiniMax HD voice synthesis with cloning\",\n    agentVisible: true,\n  },\n  \"minimax/speech-02-turbo\": {\n    id: \"minimax/speech-02-turbo\",\n    name: \"MiniMax Speech 02 Turbo\",\n    providerId: \"speech-02-turbo\",\n    provider: \"minimax\",\n    capabilities: {\n      tts: true,\n      languages: [\"multi\"],\n      voiceCloning: true,\n      streaming: true,\n    },\n    pricing: {\n      billingUnit: \"per_1k_chars\",\n      baseCost: 0.01,\n      creditCost: 1,\n    },\n    description: \"MiniMax low-latency TTS\",\n    agentVisible: true,\n  },\n\n  // ─── Deepgram STT ───────────────────────────────────────────────────────\n  // Source: https://deepgram.com/pricing — Nova-3 streaming $0.0058/min,\n  // batch $0.0048/min; Nova-2 streaming $0.0043, batch $0.0036.\n  \"deepgram/nova-3\": {\n    id: \"deepgram/nova-3\",\n    name: \"Deepgram Nova-3\",\n    providerId: \"nova-3\",\n    provider: \"deepgram\",\n    capabilities: {\n      stt: true,\n      languages: [\"multi\"],\n      timestamps: true,\n      diarization: true,\n      languageDetection: true,\n    },\n    pricing: {\n      billingUnit: \"per_minute\",\n      baseCost: 0.0048,\n      creditCost: 1,\n    },\n    description: \"Latest Deepgram model — best accuracy\",\n    agentVisible: true,\n  },\n  \"deepgram/nova-2\": {\n    id: \"deepgram/nova-2\",\n    name: \"Deepgram Nova-2\",\n    providerId: \"nova-2\",\n    provider: \"deepgram\",\n    capabilities: {\n      stt: true,\n      languages: [\"multi\"],\n      timestamps: true,\n      diarization: true,\n      languageDetection: true,\n    },\n    pricing: {\n      billingUnit: \"per_minute\",\n      baseCost: 0.0036,\n      creditCost: 1,\n    },\n    description: \"Production Deepgram model — battle-tested\",\n    agentVisible: true,\n  },\n\n  // ─── AssemblyAI STT ─────────────────────────────────────────────────────\n  // Source: https://www.assemblyai.com/pricing — Universal-2 $0.37/hour,\n  // Nano $0.12/hour. Per-minute math: /60.\n  \"assemblyai/universal-2\": {\n    id: \"assemblyai/universal-2\",\n    name: \"AssemblyAI Universal-2\",\n    providerId: \"best\",\n    provider: \"assemblyai\",\n    capabilities: {\n      stt: true,\n      languages: [\"multi\"],\n      timestamps: true,\n      diarization: true,\n      languageDetection: true,\n    },\n    pricing: {\n      billingUnit: \"per_minute\",\n      baseCost: 0.00617,\n      creditCost: 1,\n    },\n    description: \"AssemblyAI flagship — high accuracy, slow\",\n    agentVisible: true,\n  },\n  \"assemblyai/nano\": {\n    id: \"assemblyai/nano\",\n    name: \"AssemblyAI Nano\",\n    providerId: \"nano\",\n    provider: \"assemblyai\",\n    capabilities: {\n      stt: true,\n      languages: [\"multi\"],\n      timestamps: true,\n    },\n    pricing: {\n      billingUnit: \"per_minute\",\n      baseCost: 0.002,\n      creditCost: 1,\n    },\n    description: \"AssemblyAI fast tier\",\n    agentVisible: true,\n  },\n\n  // ─── Replicate-hosted STT (cheap, slow) ─────────────────────────────────\n  // Source: https://replicate.com/<owner>/<model> per-run pricing pages.\n  \"replicate/incredibly-fast-whisper\": {\n    id: \"replicate/incredibly-fast-whisper\",\n    name: \"Incredibly Fast Whisper (Replicate)\",\n    providerId: \"vaibhavs10/incredibly-fast-whisper\",\n    provider: \"replicate\",\n    capabilities: {\n      stt: true,\n      languages: [\"multi\"],\n      timestamps: true,\n      languageDetection: true,\n    },\n    pricing: {\n      billingUnit: \"per_minute\",\n      baseCost: 0.0023,\n      creditCost: 1,\n    },\n    description: \"Whisper-large-v3 on Replicate, fast-decoded\",\n    agentVisible: true,\n  },\n\n  // ─── Replicate-hosted TTS ───────────────────────────────────────────────\n  // Estimates per ~30s clip → per-1k-chars normalization (~150 chars / 30s).\n  \"replicate/chatterbox\": {\n    id: \"replicate/chatterbox\",\n    name: \"Chatterbox (Replicate)\",\n    providerId: \"resemble-ai/chatterbox\",\n    provider: \"replicate\",\n    capabilities: {\n      tts: true,\n      languages: [\"en\"],\n      voiceCloning: true,\n    },\n    pricing: {\n      billingUnit: \"per_1k_chars\",\n      baseCost: 0.02,\n      creditCost: 2,\n    },\n    description: \"Replicate-hosted voice cloning TTS\",\n    agentVisible: false,\n  },\n}\n\n// ─── Derived arrays ──────────────────────────────────────────────────────\n\nexport const AUDIO_MODEL_IDS = Object.keys(AUDIO_MODEL_CATALOG) as [\n  string,\n  ...string[],\n]\n\nexport const AGENT_AUDIO_MODEL_IDS = Object.entries(AUDIO_MODEL_CATALOG)\n  .filter(([, m]) => m.agentVisible)\n  .map(([id]) => id) as [string, ...string[]]\n\nexport const AGENT_TTS_MODEL_IDS = Object.entries(AUDIO_MODEL_CATALOG)\n  .filter(([, m]) => m.agentVisible && m.capabilities.tts)\n  .map(([id]) => id) as [string, ...string[]]\n\nexport const AGENT_STT_MODEL_IDS = Object.entries(AUDIO_MODEL_CATALOG)\n  .filter(([, m]) => m.agentVisible && m.capabilities.stt)\n  .map(([id]) => id) as [string, ...string[]]\n\n// ─── S2S / Gemini Live source of truth ───────────────────────────────────\n// These were hardcoded (and drifting) across the gemini-live adapter,\n// tts-providers, and per-app voice env. The catalog is the single source;\n// consumers should import from here instead of re-declaring the strings.\n\n/** All catalog s2s entry ids (internal billing ids). */\nexport const S2S_MODEL_IDS = Object.entries(AUDIO_MODEL_CATALOG)\n  .filter(([, m]) => m.capabilities.s2s)\n  .map(([id]) => id)\n\n/** Provider model strings the Gemini Live adapter accepts server-side,\n *  derived from the catalog's google/s2s entries. */\nexport const GEMINI_LIVE_MODEL_IDS: string[] = Object.values(\n  AUDIO_MODEL_CATALOG\n)\n  .filter(m => m.capabilities.s2s && m.provider === \"google\")\n  .map(m => m.providerId)\n\n/** Default Gemini Live model — the low-latency flash tier (`google/gemini-live`). */\nexport const DEFAULT_GEMINI_LIVE_MODEL: string =\n  AUDIO_MODEL_CATALOG[\"google/gemini-live\"]?.providerId ??\n  \"gemini-2.5-flash-preview-tts\"\n\n/** True when `id` is one of the accepted Gemini Live provider model strings. */\nexport function isAcceptedGeminiLiveModel(id: string): boolean {\n  return GEMINI_LIVE_MODEL_IDS.includes(id)\n}\n\n/** Generate a markdown table of agent-visible TTS / STT models for prompt blocks */\nexport function generateAudioModelTable(modality?: AudioModality): string {\n  return Object.values(AUDIO_MODEL_CATALOG)\n    .filter(m => m.agentVisible && (!modality || m.capabilities[modality]))\n    .map(m => {\n      const unit =\n        m.pricing.billingUnit === \"per_1k_chars\"\n          ? \"1k chars\"\n          : m.pricing.billingUnit === \"per_minute\"\n            ? \"min\"\n            : \"sec\"\n      const modalities = [\n        m.capabilities.tts ? \"TTS\" : null,\n        m.capabilities.stt ? \"STT\" : null,\n        m.capabilities.s2s ? \"S2S\" : null,\n      ]\n        .filter(Boolean)\n        .join(\"+\")\n      return `| \\`${m.id}\\` | ${modalities} | $${m.pricing.baseCost}/${unit} | ${m.description} |`\n    })\n    .join(\"\\n\")\n}\n"]}