Retell charges per minute, with add-ons such as PII removal priced per minute too. Omnia has no per-minute meter: free while you build, the core free in production, and an enterprise licence per environment per year from £30,000.

omnia vs retell ai

Retell versions and tests voice agents, and Omnia tests them the way a caller hears them, as audio.

Omnia’s voice tests run as audio, so a caller going quiet or talking over the agent is caught before release. Retell’s batch tests simulate a text conversation, and timing problems only show up on a web call someone makes by hand.

A new prompt passes every batch test.

on omnia

On Omnia it also has to pass a set of calls, including a caller who goes quiet to find a card. The new prompt starts talking again before they’re back, so the version stays at the gate. docs

on retell ai

Simulation runs as a text conversation. Voice is tested with manual web calls, and agents using a custom LLM aren’t supported in simulation. their docs

Tests that make the call

Voice scenarios run as duplex audio on every version. Each call is recorded for playback. docs

Retell AI: Automated simulation is text; voice is tested by hand. source

Rollback on a slice of calls

A version that passes the suite takes calls in steps, and Omnia moves traffic back if its metrics drop. Callers already on the line finish on the version they started with. docs enterprise

Retell AI: Published versions are read-only with prod and staging tags; rolling back means moving the tag. Automatic rollback isn’t documented. source

Tests that can’t book a real appointment

Tests run against the tools in an environment you choose, and tool-call assertions check what the agent did. Policy on every call refuses anything you haven’t allowed. docs enterprise

Retell AI: Unmocked functions call their real endpoints during simulation, and MCP tools ignore mocks. source

At a glance

Topic	Omnia	Retell AI, per its docs
Immutable versions	yes A versioned pack. docs	documented Read-only published versions. source
Batch regression tests	yes Yes, and they gate the release. docs	documented Batch tests with an API. source
Automated voice tests	yes Barge-in, overlap, dead air. docs	not documented Text simulation only. source
Adversarial callers	yes Yes. docs	not documented Not documented. source
Stepped rollout, auto rollback	yes Yes. docs	partly A/B split across agents. source
PII redaction	yes Sessions and memory. docs enterprise	documented Per-minute add-on. source
Runs in your infrastructure	yes Yes. docs	not documented Not documented. source

Common questions

How is each priced?: Retell charges per minute, with add-ons such as PII removal priced per minute too. Omnia has no per-minute meter: free while you build, the core free in production, and an enterprise licencelicense per environment per year from £30,000.

Put your own agent through the gate.

Install Omnia in your cluster and run it free while you build.

Retell AI claims are from Retell AI’s published documentation, checked 22 October 2026. Spot something out of date? Tell us.

Install it in your cluster Or walk through it with us →