Where they are alike
Both are platforms where you build the agent. You define a prompt or a node flow, connect it to a number, test it by text and audio, and see the call history with post-call analysis. Both let you transfer to a human, capture structured data and call your own system through functions pointing at your URL.
If you were looking for a feature list where one wins outright, you will not find it. The real differences are elsewhere.
Where calls are processed
This is the underlying difference. The Retell AI compliance page (docs.retellai.com/general/compliance), checked on 4 October 2026, says: «We do not currently operate services within the European Union». They comply with GDPR by relying on AWS and offer a data processing addendum with standard contractual clauses.
Lera stores calls, transcripts and recordings in the European Union. Handling a call also involves vendors headquartered outside the EU (synthetic voice or the language model, for example), some with servers in the EU and all under standard contractual clauses; the full list is on the Lera subprocessors page. For many projects it makes no difference. For a clinic, a public body, or any customer who asks you in writing where the data is stored, it stops making no difference.
How the minute is billed
Lera has no flat rate. The per-minute price comes from the agent configuration, because voice and language model cost differently. It starts at €0.073 per minute with the cheapest combination and goes up from there. Billing is per second, with no monthly fee, on prepaid balance.
The trial is 10 € of credit with no card and no deadline. It ends when you spend the credit, not when a clock runs out.
The practical consequence is that two of your own agents can cost differently depending on how you configure them, and that a quiet month costs little instead of costing the same subscription.
Where Retell AI wins
They have been at it longer and it shows. They have SOC 2 and offer a business associate agreement for HIPAA-covered workloads, which matters if you handle health data in the United States. Their integration catalogue is larger, their documentation is more extensive and their community is bigger, so it is easier to find someone who already solved your case.
If your operation is American, that maturity weighs more than European residency.
Where Lera wins
On where the data is stored and on European Spanish. The engine has its own mechanisms for Spanish (how it says times, dates, emails and references), and the voices it recommends are peninsular Spanish.
And on operational detail. Call recording ships switched off and you turn it on if you want it. When a customer clones their voice, the audio sample is not kept, only its fingerprint, because a voice sample is biometric data and keeping it without needing it is keeping the risk without the benefit.
How to choose without overthinking
If you invoice in Europe and someone will ask you where the data sits, start with residency. If your volume moves up and down, look at the billing model before the feature list. If you need HIPAA or a very broad integration ecosystem, Retell starts ahead.
And either way, try it with real calls before deciding. A voice agent is judged by listening to it, not by reading a table.
Trademark note
Retell AI is a trademark of its owner. Lera is not affiliated with or endorsed by Retell AI. The figures cited come from their public documentation (docs.retellai.com/general/compliance), checked on 4 October 2026, and may have changed since.