Retell AI pricing per minute: $0.07 to $0.31, line by line

Retell lists $0.055 for voice infrastructure, $0.015 for most TTS, an LLM line and $0.015 telephony. Add them up, and see how Sume's TTS price compares.

5 min readSume
All posts

Retell AI's pricing page gives a voice agent range of $0.07 to $0.31 per minute, built from separate lines: $0.055 per minute for voice infrastructure, $0.015 for most text-to-speech voices, an LLM line, and telephony from $0.015. A stack of infrastructure, a $0.015 voice, the GPT 5.2 line and telephony comes to $0.141 a minute (my arithmetic). Sume's text-to-speech is about $0.036 to $0.038 per minute of speech, roughly 2.4 to 2.5 times Retell's $0.015 TTS line, and Sume is a file job, not a live agent.

Retell figures are from its pricing page, read on 2026-10-03. The per-minute conversion for Sume uses Cartesia's published 750 to 800 characters per minute.

The lines Retell lists

Retell prices the pieces separately and then states a standard range. These are the numbers on the page.

Retell AI per-minute lines (read 2026-10-03)
LinePer minute
Voice agent, standard range$0.07 to $0.31
Retell voice infrastructure$0.055
TTS: Retell platform, Minimax, Fish, Cartesia, OpenAI, Inworld$0.015
TTS: ElevenLabs$0.040
LLM: GPT 5.2$0.056
LLM: GPT 5.6 Terra or Claude 5 Sonnet$0.064
LLM: GPT 5.4$0.080
LLM: GPT 5.5$0.16
LLM: budget option (GPT 5 nano)$0.0016
Telephony base rate$0.015, varies by country and provider

Adding it up

A stack of infrastructure ($0.055), a $0.015 voice, GPT 5.2 ($0.056) and telephony ($0.015) is $0.141 a minute. Swap in the ElevenLabs voice and it is $0.166. Pick GPT 5.5 and a $0.015 voice, and the same stack is $0.245. The page does not show how it builds the $0.07 and $0.31 ends of its range.

Concurrency is priced too: the page says 20 concurrent calls are included free, with additional capacity at $8.00 per concurrency per month, so 50 extra slots add $400 a month. It also says SIP trunking with custom telephony carries no charge.

Sume's TTS line next to Retell's

Sume charges $0.0475 per 1,000 characters for text-to-speech, rounded up to whole cents per request, with up to 20,000 characters or 1,200 seconds of audio per request. At 750 to 800 characters per minute of speech that is $0.0356 to $0.038. Against Retell's $0.015, the multiple is about 2.4 to 2.5.

That gap is the price of a different product. Retell's line is for speech inside a live call. Sume's line produces a stored audio file you can reuse, mix under video, or attach to an avatar clip.

TTS per minute of speech (as of 2026-10-03)
OptionPer minute of speech
Retell, most voices$0.015
Retell, ElevenLabs voice$0.040
Sume TTSabout $0.036 to $0.038

What Sume does and does not do

Sume does not place calls, take calls or run an LLM turn loop for a phone agent. It is useful around the call.

  • Render scripted lines once (greeting, hours, legal notice) and play the file, so the live meter only runs on the conversation. See pre-rendering fixed replies.
  • Create voicemail and hold-message audio in the format your phone system wants.
  • Transcribe call recordings at $0.01 a minute, in jobs of up to 600 seconds.

Monthly totals for four stacks

Each row below sums the listed lines at 10,000 minutes a month. The model choice moves the total more than the voice does.

Retell stacks at 10,000 minutes a month (read 2026-10-03)
StackPer minutePer month
Infra + $0.015 voice + $0.0016 LLM + telephony$0.0866$866
Infra + $0.015 voice + GPT 5.2 + telephony$0.141$1,410
Infra + ElevenLabs voice + GPT 5.2 + telephony$0.166$1,660
Infra + $0.015 voice + GPT 5.5 + telephony$0.245$2,450

Fixed script, repeated often

A 200-character opening line, rendered with Sume, costs one cent once. Spoken live on Retell's $0.015 line it is roughly 15 seconds of speech, about $0.004, so a one-off render pays back after three calls on cost alone. That comparison ignores that the call itself is still billed at the infrastructure rate while the file plays.

How to decide

Retell's TTS line is the cheaper way to speak inside a call. Use Sume where the same line would be spoken thousands of times: one render of a 200-character greeting is one cent. For the wider stack math see voice agent cost per minute.

Sources

Related posts

More in Pricing

All Pricing posts

Written by Sume