{
  "listings": [
    {
      "slug": "bolna",
      "name": "Bolna API + MCP",
      "company": "Voxlabs Private Limited",
      "grade": "D",
      "fit": "caveat",
      "why": "STT, LLM and TTS are chosen per agent from many providers, the core is MIT open source, executions report time_to_first_audio and a per-component cost_breakdown. Local turn detection and one-clock latency are not documented.",
      "source": "https://www.bolna.ai/docs/providers.md",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/bolna"
    },
    {
      "slug": "cuelight",
      "name": "CueLight",
      "company": "sshah03 (individual GitHub account, no name or organisation stated)",
      "grade": "D",
      "fit": "caveat",
      "why": "Supplies the local turn-detection piece under Apache-2.0, running on a fraction of one CPU core with no framework tie. English only, self-reported benchmarks, not on PyPI; latency and cost of the rest of the pipeline are out of scope.",
      "source": "https://raw.githubusercontent.com/sshah03/CueLight/main/README.md",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/cuelight"
    },
    {
      "slug": "deepgram-voice-agent",
      "name": "Deepgram Voice Agent API",
      "company": "Deepgram, Inc.",
      "grade": "BB",
      "fit": "caveat",
      "why": "A managed chain where the LLM and TTS are config fields (a BYO LLM endpoint; ElevenLabs, OpenAI, Cartesia or Polly TTS) and STT is Deepgram. Turn detection is server-side and billing is a flat per-minute tier, not a per-provider split.",
      "source": "https://developers.deepgram.com/docs/configure-voice-agent.md",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/deepgram-voice-agent"
    },
    {
      "slug": "livekit-agents",
      "name": "LiveKit Agents",
      "company": "LiveKit Incorporated",
      "grade": "BB",
      "fit": "caveat",
      "why": "Swappable STT, LLM and TTS plugins, local Silero VAD and a CPU v1-mini turn model, with per-turn ttft and ttfb metrics; but usage is reported as counts not cost, and adaptive interruption runs on LiveKit Inference, billed past an allowance when self-hosted.",
      "source": "https://docs.livekit.io/llms-full.txt",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/livekit-agents"
    },
    {
      "slug": "pipecat",
      "name": "Pipecat",
      "company": "Daily",
      "grade": "B",
      "fit": "caveat",
      "why": "Swappable STT, LLM and TTS services with local Silero VAD and bundled Smart Turn v3 weights, plus UserBotLatencyObserver; per-service usage metrics give tokens and characters, so cost per provider must be computed by the developer.",
      "source": "https://docs.pipecat.ai/llms-full.txt",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/pipecat"
    },
    {
      "slug": "pipecat-phonellm",
      "name": "Pipecat PhoneLLM",
      "company": "Daily",
      "grade": "B",
      "fit": "caveat",
      "why": "Built as the swappable LLM in a cascaded pipeline; Daily measured 17 ms TTFT on one B200 for Mini, which is a vendor figure, and Max needs dedicated GPUs with no per-token route.",
      "source": "https://huggingface.co/pipecat-ai/phonellm-mini-alpha-2",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/pipecat-phonellm"
    },
    {
      "slug": "vaani-voice",
      "name": "Vaani Developer Platform",
      "company": "Vaani AI Research Private Limited (Bengaluru, India, CIN U62099KA2025PTC201212)",
      "grade": "E",
      "fit": "caveat",
      "why": "STT, LLM and TTS (ears, brain, mouth) are chosen per agent from Deepgram, AssemblyAI, Google, Azure, Cartesia, ElevenLabs, OpenAI and others, BYOL is supported, and the dashboard estimates per-minute cost by stage. Hosted turn handling and no per-turn latency metrics.",
      "source": "https://docs.vaanivoice.ai/api-reference/update-persona",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/vaani-voice"
    },
    {
      "slug": "vapi",
      "name": "Vapi API + MCP",
      "company": "Vapi Inc.",
      "grade": "B",
      "fit": "caveat",
      "why": "Transcriber, model and voice are chosen per assistant from a provider list, with per-turn TurnLatency fields and a per-call costBreakdown by stt, llm, tts and transport. Turn detection runs on Vapi's servers, not locally. An independent test measured 536 ms average, P95 up to 1.2 s.",
      "source": "https://docs.vapi.ai/openapi/api-reference.json",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/vapi"
    },
    {
      "slug": "vocily",
      "name": "Vocily AI",
      "company": "VOCILY AI PRIVATE LIMITED",
      "grade": "C",
      "fit": "caveat",
      "why": "Transcriber, model and voice are picked from catalogues with backup providers, but turn handling is hosted, no latency metrics are exposed, and call cost is a total with a per-minute rate rather than a per-provider split.",
      "source": "https://docs.vocily.ai/developers/openapi.json",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/vocily"
    },
    {
      "slug": "hume-evi",
      "name": "Hume EVI (Empathic Voice Interface)",
      "company": "Hume AI, Inc.",
      "grade": "C",
      "fit": "unchecked",
      "why": "EVI docs withdrawn: every EVI page, including the overview and tool-use guide, redirects to the Audio Decisions overview, and the EVI AsyncAPI file returns 404. The last assessment recorded EVI access ending 13 November 2026.",
      "source": "https://dev.hume.ai/docs/speech-to-speech-evi/overview",
      "confidence": "low",
      "url": "https://www.anchorterminal.com/tools/hume-evi"
    },
    {
      "slug": "amazon-nova-sonic",
      "name": "Amazon Nova 2 Sonic",
      "company": "Amazon Web Services, Inc.",
      "grade": "B",
      "fit": "no",
      "why": "One speech-to-speech model with server-side endpointing (1.5 to 2 second pauses); no swappable STT or TTS and no local turn detection.",
      "source": "https://docs.aws.amazon.com/nova/latest/nova2-userguide/sonic-turn-taking.md",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/amazon-nova-sonic"
    },
    {
      "slug": "assemblyai-voice-agent",
      "name": "AssemblyAI Voice Agent API",
      "company": "AssemblyAI, Inc.",
      "grade": "C",
      "fit": "no",
      "why": "A managed pipeline with AssemblyAI's own STT and voices; only the LLM can be swapped. Turn detection is server-side and the rate is flat, so no per-provider cost.",
      "source": "https://www.assemblyai.com/docs/voice-agents/voice-agent-api.md",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/assemblyai-voice-agent"
    },
    {
      "slug": "bland-ai",
      "name": "Bland AI API + MCP",
      "company": "Intelliga Corporation",
      "grade": "B",
      "fit": "no",
      "why": "Bland runs its own STT, LLM, TTS and telephony as one bundled per-minute product; no provider swap or local turn detection.",
      "source": "https://docs.bland.ai/llms.txt",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/bland-ai"
    },
    {
      "slug": "elevenlabs-agents",
      "name": "ElevenLabs Agents API + MCP",
      "company": "Eleven Labs Inc.",
      "grade": "BB",
      "fit": "no",
      "why": "A hosted pipeline: the LLM or a custom LLM can be swapped, but ASR, TTS and turn-taking are ElevenLabs'. One account self-hosted instead to get about a tenth of the cost.",
      "source": "https://elevenlabs.io/docs/eleven-agents/llms.txt",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/elevenlabs-agents"
    },
    {
      "slug": "fone-me",
      "name": "fone.me",
      "company": "Fone",
      "grade": "F",
      "fit": "no",
      "why": "A finished consumer product; no pipeline, provider choice or latency metrics are exposed.",
      "source": "https://www.fone.me/llms.txt",
      "confidence": "low",
      "url": "https://www.anchorterminal.com/tools/fone-me"
    },
    {
      "slug": "fresha-ai-concierge",
      "name": "Fresha AI Concierge",
      "company": "Fresha.com SV Ltd",
      "grade": "F",
      "fit": "no",
      "why": "A closed add-on inside Fresha; voice, accent and tone are chosen in settings but no pipeline or provider is exposed.",
      "source": "https://www.fresha.com/for-business/features/ai-concierge",
      "confidence": "low",
      "url": "https://www.anchorterminal.com/tools/fresha-ai-concierge"
    },
    {
      "slug": "gemini-live",
      "name": "Gemini Live API",
      "company": "Google LLC",
      "grade": "C",
      "fit": "no",
      "why": "One speech-to-speech model; STT, LLM and TTS cannot be swapped and no per-turn latency metrics are exposed.",
      "source": "https://aistudio.google.com/docs/live-api/capabilities.md",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/gemini-live"
    },
    {
      "slug": "gpt-live",
      "name": "GPT-Live",
      "company": "OpenAI OpCo, LLC",
      "grade": "BB",
      "fit": "no",
      "why": "The voice layer is gpt-live-1 only; the backend model can be chosen, but STT and TTS cannot be swapped and no local turn detection is offered.",
      "source": "https://developers.openai.com/api/docs/models/gpt-live-1.md",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/gpt-live"
    },
    {
      "slug": "harmony-ai",
      "name": "Harmony.ai",
      "company": "monday.com Ltd.",
      "grade": "E",
      "fit": "no",
      "why": "A closed managed product on the vendor's own model; no provider choice, latency metrics or pipeline access.",
      "source": "https://harmony.ai/platform/ai-dialer",
      "confidence": "low",
      "url": "https://www.anchorterminal.com/tools/harmony-ai"
    },
    {
      "slug": "inworld-realtime",
      "name": "Inworld Realtime API",
      "company": "Theai, Inc. dba Inworld AI",
      "grade": "B",
      "fit": "no",
      "why": "A managed STT, LLM and TTS chain where only the LLM is chosen per session; STT and TTS are Inworld's, turn detection runs server-side, and no per-turn latency metrics on one clock are exposed.",
      "source": "https://docs.inworld.ai/realtime/usage/using-realtime-models.md",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/inworld-realtime"
    },
    {
      "slug": "openai-realtime",
      "name": "OpenAI Realtime API",
      "company": "OpenAI OpCo, LLC",
      "grade": "BB",
      "fit": "no",
      "why": "One speech-to-speech model; STT, LLM and TTS cannot be swapped, turn detection runs on OpenAI's server, and cost is per token rather than per provider.",
      "source": "https://developers.openai.com/api/docs/guides/realtime-conversations.md",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/openai-realtime"
    },
    {
      "slug": "placecall",
      "name": "PlaceCall",
      "company": "VOYGR Tech, Inc.",
      "grade": "C",
      "fit": "no",
      "why": "A managed calling service; no choice of STT, LLM or TTS and no latency metrics.",
      "source": "https://github.com/voygr-tech/placecall/blob/main/README.md",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/placecall"
    },
    {
      "slug": "retell-ai",
      "name": "Retell AI API + MCP",
      "company": "Retell AI Inc.",
      "grade": "B",
      "fit": "no",
      "why": "A hosted platform: LLM and voice come from Retell's menu or a custom LLM socket, turn detection runs on Retell's servers, so the caller-built pipeline with local interruption is not offered.",
      "source": "https://docs.retellai.com/llms-full.txt",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/retell-ai"
    },
    {
      "slug": "rime-prism",
      "name": "Rime Prism",
      "company": "Rime Labs, Inc.",
      "grade": "B",
      "fit": "no",
      "why": "One speech-to-speech model on Rime's cloud; no swappable STT, LLM or TTS and turn detection is server-side.",
      "source": "https://docs.rime.ai/docs/prism/overview.md",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/rime-prism"
    },
    {
      "slug": "synthflow",
      "name": "Synthflow API + MCP",
      "company": "AgentFlow AI GmbH",
      "grade": "D",
      "fit": "no",
      "why": "A managed platform, not a pipeline you assemble: non-enterprise accounts get only Synthflow LLM, turn handling runs on Synthflow, and no per-turn latency or per-provider cost breakdown was found.",
      "source": "https://docs.synthflow.ai/general-configuration.md",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/synthflow"
    },
    {
      "slug": "ultravox",
      "name": "Ultravox Realtime API",
      "company": "Theai, Inc. dba Inworld AI",
      "grade": "C",
      "fit": "no",
      "why": "A hosted speech-native model with a TTS voice; STT and LLM cannot be swapped, turn detection is server-side VAD, and no per-turn latency metrics were found in the schema.",
      "source": "https://api.ultravox.ai/api/schema/",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/ultravox"
    },
    {
      "slug": "vogent",
      "name": "Vogent API",
      "company": "Monoid, Inc.",
      "grade": "E",
      "fit": "no",
      "why": "A hosted phone platform: models are chosen from Vogent's list and turn detection runs server-side, with no per-turn latency metrics or per-provider cost in the API.",
      "source": "https://docs.vogent.ai/api-reference/openapi.yaml",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/vogent"
    }
  ],
  "marks": "yes: the docs show it does the job as the success test says. caveat: it does the job with a limit an agent should know. no: it doesn't do this job. unchecked: we couldn't read it.",
  "meta": {
    "attribution": "Anchor Terminal (https://www.anchorterminal.com)",
    "attributionRequired": true,
    "credit": "CC BY 4.0, credit Anchor Terminal and link the page or API address you took it from (https://www.anchorterminal.com/terms/#data)",
    "docs": "https://www.anchorterminal.com/docs/",
    "generatedAt": "2026-11-16",
    "license": "CC-BY-4.0",
    "method": "https://www.anchorterminal.com/benchmark/",
    "methodology": "0.8",
    "openapi": "https://www.anchorterminal.com/openapi.json",
    "preview": false,
    "run": "2026-10-01",
    "runLabel": "October 2026 research run",
    "terms": "https://www.anchorterminal.com/terms/#data"
  },
  "useCase": {
    "accounts": 5,
    "after": [
      "voice-agents.inbound-receptionist",
      "hosting.deploy"
    ],
    "area": "voice",
    "asks": [
      "build a voice agent cheaper than ElevenLabs",
      "get voice-to-voice under 800 ms",
      "swap the LLM without changing the rest"
    ],
    "before": [
      "inference.llm",
      "speech.stt",
      "speech.tts"
    ],
    "blockers": [
      "turn-detector model licence forbids use outside one framework",
      "GPU capacity needed to self-host the speech models",
      "latency figures measured on two clocks are wrong"
    ],
    "capabilities": [
      "voice.pipeline",
      "voice.agent",
      "voice.tools",
      "agent.mcp-client"
    ],
    "category": "voice-agents",
    "categoryName": "Conversational voice agents",
    "checked": "2026-11-12",
    "companies": {
      "fit": 0,
      "fitWithLimit": 8,
      "tagged": 113
    },
    "competition": "none",
    "fits": null,
    "fitsWithLimit": [
      "bolna",
      "cuelight",
      "deepgram-voice-agent",
      "livekit-agents",
      "pipecat",
      "pipecat-phonellm",
      "vaani-voice",
      "vapi",
      "vocily"
    ],
    "id": "voice-agents.low-latency-pipeline-build",
    "listings": {
      "caveat": 9,
      "no": 17,
      "notRead": 0,
      "tagged": 130,
      "unchecked": 1,
      "yes": 0
    },
    "name": "Assemble your own STT, LLM and TTS pipeline and tune latency and cost",
    "status": "draft",
    "success": "Interruption detected locally without a network round trip, and the per-minute cost broken down by provider. Measured: median voice-to-voice latency under 800 ms on one clock.",
    "url": "https://www.anchorterminal.com/api/v1/use-cases/voice-agents.low-latency-pipeline-build.json",
    "weight": 10,
    "weightBasis": "5 of 31 accounts (acc-01, acc-08, acc-18, acc-21, acc-28)"
  }
}
