Retell AI

What it does

Writes tests

PartlyProduces graded conversation test cases for its own agents, not software tests.Run graded test cases to catch regressions before launch.docs.retellai.com/general/introduction ↗

Done-for-you service

PartlyYou run the software yourself in Retell's no-code studio; hands-on implementation by Retell's people is sold only as an Enterprise add-on, not as a managed service.Dedicated Implementation Support (add-on)retellai.com/pricing ↗

How much it does unattended

Runs autonomously

YesAgents take and place whole calls without a person on the line, including scheduled outbound campaigns with pacing, retries and voicemail handling; a human can listen in and take over by choice, not by requirement.Upload your audience, schedule campaigns, control pacing and retries, handle voicemail, and let AI agents call at scale.retellai.com ↗

Multi-agent

YesOne call can be handed between several specialised AI agents mid-conversation, with the destination agent inheriting the full transcript, so work is split across agents rather than one monolith.Agent Transfer (also known as Agent Swap) enables you to modularize tasks and re-use specialized agents without relying on traditional phone-based…docs.retellai.com/build/conversation-fl… ↗

Agent permissions

PartlyProhibited-topic guardrails screen both the agent's replies and user input, and tools are explicitly wired per node, but guardrails only swap in a placeholder message rather than stopping or escalating the call.Guardrails do not end the call, transfer the call, or trigger any other action — they only replace the problematic message.docs.retellai.com/build/guardrails ↗

Which models it runs on

Claude

YesClaude models are selectable as the agent's conversation model and priced per minute for voice and per message for chat, alongside GPT and Gemini options.Claude 5 SonnetRecommended | $0.064/minuteretellai.com/pricing ↗

GPT

YesMultiple GPT tiers are offered as conversation models, each with its own per-minute rate and an optional faster tier for lower latency.GPT 5.4Recommended | $0.080/minute | $0.16/minuteretellai.com/pricing ↗

Open models

PartlyRetell's own model menu is limited to hosted GPT, Claude and Gemini, but the custom-LLM path lets you put a self-hosted open-weights model behind the agent over a WebSocket you operate.A fine-tuned model, a self-hosted open-weights model, or a provider outside the supported options.docs.retellai.com/integrate-llm/overview ↗

Model choice

YesYou pick the conversation model per agent from a published, per-minute-priced list, and you can substitute your own model entirely through the custom-LLM WebSocket.With a custom LLM, you own response generation.docs.retellai.com/integrate-llm/overview ↗

Where you use it

In your editor

PartlyThere is no editor plugin for building calls, but the hosted MCP server lets IDE-based assistants such as Cursor create, publish and inspect Retell agents from inside the editor.Retell supports the Model Context Protocol (MCP) so you can build Retell AI voice agents directly from MCP-capable clients (Cursor, Claude Desktop,…docs.retellai.com/get-started/mcp-server ↗

On the command line

NoThe docs enumerate every way to drive Retell: dashboard, REST API, Node/Python SDKs and MCP server. No command-line tool is among them.Build entirely from the dashboard with no code, or go fully programmatic with the API, SDKs, and an MCP server.docs.retellai.com/general/introduction ↗

In a browser

YesEverything is built and operated in a hosted browser dashboard, and agents themselves can also be embedded on your own site as a web call or chat widget.Embed the Retell website widget on your site with a single script tag — supports chat, voice, and callback modes powered by your chat and voice…docs.retellai.com/deploy/chat-widget ↗

Whose machine it runs on

Self-hosted

NoThe platform runs on Retell's own infrastructure: pay-as-you-go uses shared servers and the most isolation offered is a dedicated Retell server sold as an Enterprise add-on, with no customer-infrastructure deployment documented.Server | Shared | Dedicated stable server (add-on)retellai.com/pricing ↗

Not established: Open source

What it costs to run

How it meters

YesBilling is consumption-based: voice agents bill per call minute split across voice infra, model and telephony, chat agents bill per AI message, with named add-ons metered per minute or per dial.$0.07-$0.31 / min for AI Voice Agentsretellai.com/pricing ↗

Free tier

PartlyYou can sign up and run real calls without talking to sales, but the free allowance is a one-off $10 trial credit per email address rather than a standing free plan.New accounts start with $10 in free trial credits so you can test Retell before adding a payment method. The trial credit is granted once per email…docs.retellai.com/accounts/billing ↗

API access

YesA documented REST API with Node.js and Python SDKs covers agents, calls, phone numbers and chats, and it is listed as included on the entry pay-as-you-go plan.Webhooks and API Accessretellai.com/pricing ↗

MCP server

YesRetell runs a hosted MCP endpoint authenticated with an API key, exposing agent, call, knowledge-base and QA tools so another assistant can drive the account.Retell MCP server URL: https://mcp.retellai.comdocs.retellai.com/get-started/mcp-server ↗

Bring your own key

PartlyYou can bring your own model server and your own telephony carrier over SIP, but there is no documented way to supply your own OpenAI or Anthropic keys for Retell's built-in models.Run your own LLM behind a Retell voice agent. Retell handles telephony, transcription, and turn-taking while your WebSocket server generates each…docs.retellai.com/integrate-llm/overview ↗

Buying it for a team

A company can buy it

YesWork is organised in workspaces with invited members and per-workspace billing, and an Enterprise plan adds negotiated contracts for organisations buying centrally.Billing is scoped per workspace, not per account. Every workspace, including new ones you create later, has its own payment method, credit balance,…docs.retellai.com/accounts/billing ↗

Seat model

NoNothing is charged per user: the published rate card meters minutes, messages and monthly items like phone numbers and concurrency, so team size does not change the bill.True pay as you go. Only pay for the minutes you use.retellai.com/pricing ↗

Pooled budget

YesA workspace holds one prepaid credit balance that every agent and member draws down in real time, and it blocks all new calls when it empties.When your credit balance reaches zero, new calls are blocked until you buy credits or auto recharge tops up your balance.docs.retellai.com/accounts/billing ↗

Admin controls

PartlyAdmin, Developer and Member roles constrain who can edit agents, see raw transcripts or change limits, but the pricing comparison marks role-based access control as an Enterprise capability rather than a pay-as-you-go one.To keep your workspace safe, we have an RBAC (Role-Based Access Control) system.docs.retellai.com/accounts/access-contr… ↗

Audit log

PartlyAgent configuration changes are recorded as immutable published versions with browsable history and side-by-side comparison; no workspace-level log of administrative actions such as role changes or key rotation is documented.Version control & history: you can create multiple draft versions from any past version, track your version history, and publish any draft when it's…docs.retellai.com/agent/version ↗

Single sign-on

PartlyOrdinary accounts can sign in with Google, while configurable enterprise single sign-on is listed only under the Enterprise plan and the protocol supported is not named.Custom SSOretellai.com/pricing ↗

What happens to your code

Opt out of training

PartlyPer-agent data storage settings let you keep transcripts and recordings out of Retell entirely, but the privacy policy states collected information is used to train its models and publishes no separate training opt-out.to train the artificial intelligence models that support our Services or that support our business and administrative functions;retellai.com/legal/privacy-policy ↗

Data residency

PartlyProcessing is in the United States on AWS with SCCs for EU transfers; the vendor states plainly that it runs no services inside the European Union, so EU-resident hosting is not on offer.However, please note that we do not currently operate services within the European Union.docs.retellai.com/general/compliance ↗

Getting out

YesFiltered call and chat history exports to CSV including transcripts, and list/get APIs cover the same data; purchased credits never expire but are non-refundable once bought.Open Actions, select Export, and choose the columns to include. Export-only fields include transcript data and PII-scrubbed transcripts.docs.retellai.com/features/session-hist… ↗

Certifications

YesHIPAA, GDPR and SOC 2 Type 1 and Type 2 are claimed, with BAA and DPA self-signing at no extra fee and reports available through a trust centre.we are proud to be HIPAA compliant, GDPR compliant, SOC 2 Type 1 & Type 2 certifieddocs.retellai.com/general/compliance ↗