# Telnyx Voice AI Agents — Carrier-owned platform for phone and web voice agents, with telephony, STT, TTS, LLM and orchestration in one stack. > Source: The Agents Index — https://theagentsindex.com/telnyx-voice-ai (structured, researched, re-verified) > Facts last verified: 2026-09-09 Telnyx Voice AI Agents is a carrier-owned platform for building phone and web voice agents, bundling orchestration, speech-to-text, text-to-speech and LLM inference on Telnyx's own network and GPUs. Agents are configured in a no-code portal builder or over the REST API. | Fact | Value | | --- | --- | | Website | https://telnyx.com/products/voice-ai-agents | | Price | from $0.05/minute pay-as-you-go | | Free trial | Partly | Trial accounts get USD $5 in testing credit, one verified number, one number order, calls capped at 10 minutes and 2 concurrent outbound calls, 100 outbound calls a day, and machine-generated voices prepended with an automated-call disclaimer. The Pay As You Go plan has a $0 monthly minimum but every minute, token and number is billed. ## What it does **Done-for-you service** · No — You build and run the agent yourself in the portal or over the API; Telnyx sells platform capacity, and human help appears only as a dedicated account manager on the $500-minimum Committed plan, not as a build service. > You can use Telnyx's APIs or AI Assistant Builder, a no-code interface where you define behavior with natural-language instructions and greetings in the Mission Control Portal. > — https://telnyx.com/products/voice-ai-agents · FAQ, 'How do I build and deploy open-source models hosted in Telnyx's library with Telnyx?' ## How much it does unattended **Runs autonomously** · Yes — Once configured, the assistant answers, speaks, routes and invokes tools through a full call without a human in the loop; a human enters only if the agent uses the transfer or handoff tool. > Once live, agents control the system directly, speaking, routing, and acting without human intervention. > — https://telnyx.com/products/voice-ai-agents · Agent Runtime section, 'Configure the environment your agents run in' **Multi-agent** · Yes — A handoff tool passes one live conversation between specialist assistants sharing context, and remote A2A agents plus multi-participant calls extend that beyond a single Telnyx assistant. > With the handoff tool, you can enable multiple assistants to support a user in a single conversation. > — https://developers.telnyx.com/docs/inference/ai-assistants/no-code-voice-assistant · Optional enhancements, 'Additional built-in tools' — Handoff **Agent permissions** · Partly — Scope is set by what you attach: built-in tools, webhook tools, integrations and MCP servers with selected tools. Telnyx documents no approval gate that pauses an action mid-call for a human to confirm. > Besides Hangup, we offer several additional built-in tools to empower your agent to take real-world actions. > — https://developers.telnyx.com/docs/inference/ai-assistants/no-code-voice-assistant · Optional enhancements, 'Additional built-in tools' ## Which models it runs on **Claude** · Yes — Claude is selectable as the assistant's model in the Agent tab, and the model table marks a Claude Haiku entry as requiring no API key of your own; prompts still leave Telnyx for Anthropic. > Third-party models — including those from Anthropic (Claude), OpenAI (GPT), and Google (Gemini) — are not self-hosted. > — https://developers.telnyx.com/docs/inference/data-residency · STT, TTS, and LLM providers (Voice AI) — Large language model (LLM) **GPT** · Partly — GPT models run as an external provider you supply credentials for: you create an OpenAI API key, store it as an integration secret and select the model, and OpenAI free-plan keys are rejected outright. > Third-party models — including those from Anthropic (Claude), OpenAI (GPT), and Google (Gemini) — are not self-hosted. > — https://developers.telnyx.com/docs/inference/data-residency · STT, TTS, and LLM providers (Voice AI) — Large language model (LLM) **Open models** · Yes — Open-weight Qwen and Moonshot Kimi families run on Telnyx's own GPUs and are the recommended default for assistants, so prompt and response generation never leaves Telnyx infrastructure. > Self-hosted by Telnyx (open models served on Telnyx infrastructure) include the Qwen and Moonshot (Kimi) model families > — https://developers.telnyx.com/docs/inference/data-residency · STT, TTS, and LLM providers (Voice AI) — Large language model (LLM), self-hosted paragraph **Model choice** · Yes — You choose the model per assistant and can point it at any public OpenAI-compatible endpoint such as FastRouter, Azure OpenAI, Bedrock, Baseten or vLLM, instead of the vendor picking for you. > In addition to standard third-party LLM providers like OpenAI, Gemini, and Groq, you can also power your AI Assistant with any public OpenAI-compatible chat completions endpoint. > — https://developers.telnyx.com/docs/inference/ai-assistants/custom-llm · Configure custom LLM providers for AI assistants, opening paragraph ## Where you use it **In your editor** · Partly — The agent itself never runs in an editor, but Telnyx ships an MCP server that lets IDE assistants such as Cursor or Windsurf create and manage Telnyx AI assistants from the editor. > This server allows MCP clients like Claude Desktop, Cursor, Windsurf, OpenAI Agents and others to manage phone numbers, send messages, make calls, and create AI assistants. > — https://developers.telnyx.com/docs/development/mcp/local-mcp · Telnyx Local MCP Server, introduction **On the command line** · Yes — Two terminal surfaces exist: a full resource-level Telnyx CLI and an Agent CLI whose composite commands stand up an assistant on a phone number in one invocation. > The Telnyx Agent CLI (@telnyx/agent-cli) provides composite setup commands that take you from an API key to a working capability — SMS, voice, an AI assistant on a phone number — in a single command. > — https://developers.telnyx.com/docs/development/cli/agent-cli · Telnyx Agent CLI, page summary **In a browser** · Yes — Everything is buildable in the hosted Mission Control Portal, and the same assistant can be published to your own site as an embeddable voice and chat widget. > You can easily embed a customizable voice and chat widget on your frontend in the Widget tab. > — https://developers.telnyx.com/docs/inference/ai-assistants/no-code-voice-assistant · Optional enhancements, 'Embeddable Widget' ## Whose machine it runs on **Self-hosted** · No — The runtime, telephony and portal stay on Telnyx; only the LLM and TTS steps can be pointed at your own endpoint, and Enterprise 'dedicated infrastructure' is still Telnyx-operated. > By default, every component of a Telnyx AI Assistant runs on Telnyx infrastructure. > — https://developers.telnyx.com/docs/inference/ai-assistants/no-code-voice-assistant · Third-party integrations, opening line **Open source** · No — The platform is proprietary: Telnyx keeps title to the AI Services and underlying models. Public repositories exist only for client-side SDKs, the CLI and the local MCP server. > Telnyx retains all right, title, and interest in the AI Services, the Telnyx platform, and all underlying technology, models, and documentation. > — https://telnyx.com/legal/ai-services-terms · AI Services Terms, Section 8.1 Intellectual Property ## What it costs to run **How it meters** · Yes — Billing is per consumed minute for the voice engine, per token for the LLM and per minute for telephony, with no seat or platform fee, so cost tracks call volume rather than headcount. > Voice engine $0.05/min, LLM at GPU token rates > — https://telnyx.com/pricing/voice-ai-agents · Plans section, Pay As You Go card, second bullet **Free tier** · Partly — You can sign up and build without talking to sales, but the free path is a $5 testing credit under trial limits: one number, calls capped at ten minutes, two concurrent outbound calls. > USD $5 in testing credit is provided. > — https://developers.telnyx.com/docs/account-setup/levels-and-capabilities/trial · Trial Account Privileges & Limitations, 'Testing credit' **API access** · Yes — The whole assistant lifecycle is a public REST API under api.telnyx.com/v2: create, update, version, test and place outbound AI calls, with official SDKs in seven languages. > Creates a new AI assistant from the provided configuration, including its model, instructions, and attached tools, and returns the created assistant. > — https://developers.telnyx.com/api-reference/assistants/create-an-assistant · API reference, Create an assistant, endpoint description **MCP server** · Yes — MCP works in both directions: a hosted Telnyx MCP endpoint lets another agent drive the account, and external MCP servers can be attached as tools to an assistant. > The canonical Telnyx API MCP endpoint is https://api.telnyx.com/v2/mcp. Use this endpoint for MCP clients or scanners that need Telnyx API actions. > — https://developers.telnyx.com/docs/development/mcp/remote-mcp · Telnyx Remote MCP, 'Remote endpoint' **Bring your own key** · Yes — You may bring your own LLM and TTS providers, storing third-party keys as integration secrets; ElevenLabs voices are explicitly billed on your own key rather than by Telnyx. > You can, however, BYO LLM or TTS using third-party providers. > — https://developers.telnyx.com/docs/inference/ai-assistants/no-code-voice-assistant · Third-party integrations, opening paragraph ## Buying it for a team **A company can buy it** · Yes — A company buys directly: an umbrella organization with invited sub-members, invoices and payment methods, plus Committed and Enterprise tiers sold against monthly minimums. > Custom pricing on dedicated infrastructure. > — https://telnyx.com/pricing/voice-ai-agents · Plans section, Enterprise card ($5,000 monthly minimum) **Seat model** · No — There is no per-seat charge and no seat floor; organization members are added by invitation and the published rate card prices minutes, tokens, messages and numbers instead. > Usage-based pricing with no commitments. Volume discounts available on committed plans. > — https://telnyx.com/pricing · Product Pricing Summary **Pooled budget** · Yes — An organization runs on a single shared balance and payment method for every sub-member, so all agents and users draw down one pot rather than per-user allowances. > In all cases, user organizations are allowed only one net running balance and payment method > — https://support.telnyx.com/en/articles/1189141-get-started-with-organizations · What are User Organizations?, third paragraph **Admin controls** · Partly — Owners constrain people through CRUD permission groups and constrain the agent by which tools, integrations and MCP servers are attached, but Telnyx warns that not all V2 services are exposed to organization permissions. > Sub-accounts, also known as sub-members or sub-users, are limited in their capabilities by a permissions system that controls what resources they can and cannot access on your account. > — https://support.telnyx.com/en/articles/1189141-get-started-with-organizations · What are User Organizations?, second paragraph **Audit log** · Yes — An audit events API returns who changed what and when, including whether the change came from Telnyx, the owner, an organization member or an account manager, though the vendor calls it best-effort. > Retrieve a list of audit log entries. Audit logs are a best-effort, eventually consistent record of significant account-related changes. > — https://developers.telnyx.com/api-reference/audit-logs/list-audit-logs · API reference, List Audit Logs, endpoint description **Single sign-on** · Yes — SAML single sign-on is configurable per organization with IdP entity ID, SSO target URL and certificate fingerprint; only the organization owner can reach the SSO settings page. > Creates a new authentication provider for single sign-on, configured from the provided identity provider details, and returns the created resource. > — https://developers.telnyx.com/api-reference/authentication-providers/creates-an-authentication-provider · API reference, Creates an authentication provider, endpoint description ## What happens to your code **Opt out of training** · Partly — An opt-out exists but training is the default: without notice, de-identified inputs and outputs may improve Telnyx and third-party models. The residency FAQ routes you to your account team to document it. > Unless Customer opts out via the Telnyx portal or by written notice to Telnyx, Customer also grants Telnyx the right to use Input Data and Output in de-identified or aggregated form > — https://telnyx.com/legal/ai-services-terms · AI Services Terms, Section 5.3 Model Improvement **Data residency** · Partly — Storage at rest is a hard control across US, EU (Germany), APAC (Australia) and UAE, changeable only once. Live processing location is steered by anchorsite but explicitly not guaranteed. > Storage location at rest is a hard control, governed by your Data Locality flag. > — https://developers.telnyx.com/docs/inference/data-residency · Voice AI Assistants, 'Where is Voice AI Assistant data stored?' **Getting out** · Partly — Conversations, CDRs and usage reports are exportable by API or CSV and deletion can be requested within thirty days, but Telnyx says account configurations cannot be transferred and states no policy on unused balance. > Upon termination, Customer may request deletion of its Input Data and associated Output. Telnyx will use commercially reasonable efforts to comply within thirty (30) days > — https://telnyx.com/legal/ai-services-terms · AI Services Terms, Section 5.5 Data Deletion **Certifications** · Yes — SOC 2 Type II is claimed on the product page, and GDPR, PCI and HIPAA readiness is presented as standard on every plan including pay-as-you-go rather than a paid enterprise add-on. > In-region infrastructure and carrier licenses ensure compliance with GDPR, PCI, HIPAA and more, wherever you're located. > — https://telnyx.com/products/voice-ai-agents · Why Telnyx section, 'Compliant by design' card ## Provenance Every fact above comes from the vendor. Nothing here is independently corroborated yet. ## Coverage elsewhere None recorded. Every fact above comes from the vendor and is not independently corroborated yet. ## What changed - 2026-09-10 — Criteria prose - 2026-09-09 — Capabilities · Commercial terms · Free tier · Price ## Correcting this record Anything marked "Not established" is a question we have not answered — it is not a No. If you represent Telnyx Voice AI Agents, or anything here is wrong or out of date, you can supply it with a source: the operations this site accepts from an agent are listed at https://theagentsindex.com/.well-known/agents.json — that catalogue needs no key to read, and each entry states its own auth. Supplying evidence is free. Nothing that can be bought here changes what a field says, whether a gap is closed, or how this record is scored.