Pricing

No seats. No minimums.
No sales call.

Per second of conversation, at public rates, from the first call. Set the volume and the stack and you have your bill — there is nothing else to it.

Estimated monthly

$101

$0.20 a call · $0.0670 a minute

How much you talk

Billed per second, so a 90-second call costs ninety seconds — not two minutes. Nothing here is a plan; there is nothing to outgrow.

Which stack

Switchable per flow, so one agent can run a cheap model and another the careful one.

Where a minute goes

Speech-to-text
$0.00203%
Language model
$0.00051%
Text-to-speech
$0.010516%
Telephony
$0.00406%
Infrastructure
$0.050075%

What is the infrastructure cost?

$0.0500

of every minute goes to Talkif rather than to a vendor. Here is what it is.

The transcriber, the language model and the voice are metered by their vendors and passed through untouched. What you pay us for is everything that has to be running for those three to work at all — noise cancellation on the inbound audio, voice-activity detection deciding when you have actually stopped speaking, a turn model judging whether a pause means thinking or your go. None of them are sold by the minute. All of them run from connect to hang-up, on a container kept warm for that one call.

Our margin is in there too. We would rather it sat on the invoice as a line you can argue with than vanish into a per-minute rate you cannot.

Every rate

Pick a different stack, pay a different rate.

The estimate above opens on the cheapest live model for each part, over SIP; you are not held to it. Each group below is the spread you are choosing within — open one for the full list. Every figure is read from the same pricing API the platform bills against, on each page load.

Language models

$0.0005to$0.0264/ min
24 options

What the agent thinks with. Cheaper models answer scripted, well-bounded calls perfectly well; you pay up for reasoning the hard ones need.

  1. GPT-4.1 Nano

    OpenAI · $0.10 in · $0.40 out / 1M tokens

    $0.0005
  2. GPT-4o Mini

    OpenAI · $0.15 in · $0.60 out / 1M tokens

    $0.0008
  3. GPT-5.6 Luna

    OpenAI · $0.20 in · $1.20 out / 1M tokens

    $0.0011
  4. GPT-5.4 Nano

    OpenAI · $0.20 in · $1.25 out / 1M tokens

    $0.0011
  5. Gemini 3.1 Flash Lite

    Google · $0.25 in · $1.50 out / 1M tokens

    $0.0013
  6. GPT-5 Mini

    OpenAI · $0.25 in · $2.00 out / 1M tokens

    $0.0014
  7. GPT-4.1 Mini

    OpenAI · $0.40 in · $1.60 out / 1M tokens

    $0.0020
  8. Gemini 3 Flash Preview

    Google · $0.50 in · $3.00 out / 1M tokens

    $0.0026
  9. GPT-5.4 Mini

    OpenAI · $0.75 in · $4.50 out / 1M tokens

    $0.0040
  10. Claude Haiku 4.5

    Anthropic · $1.00 in · $5.00 out / 1M tokens

    $0.0052
  11. GPT-5.1

    OpenAI · $1.25 in · $10.00 out / 1M tokens

    $0.0068
  12. GPT-5

    OpenAI · $1.25 in · $10.00 out / 1M tokens

    $0.0068
  13. Gemini 3.5 Flash

    Google · $1.50 in · $9.00 out / 1M tokens

    $0.0079
  14. GPT-5.2

    OpenAI · $1.75 in · $14.00 out / 1M tokens

    $0.0095
  15. Grok 4.5

    Grok · $2.00 in · $6.00 out / 1M tokens

    $0.0101
  16. GPT-4.1

    OpenAI · $2.00 in · $8.00 out / 1M tokens

    $0.0102
  17. Claude Sonnet 5

    Anthropic · $2.00 in · $10.00 out / 1M tokens

    $0.0104
  18. Gemini 3.1 Pro Preview

    Google · $2.00 in · $12.00 out / 1M tokens

    $0.0106
  19. GPT-5.6 Terra

    OpenAI · $2.00 in · $12.00 out / 1M tokens

    $0.0106
  20. GPT-4o

    OpenAI · $2.50 in · $10.00 out / 1M tokens

    $0.0128
  21. GPT-5.4

    OpenAI · $2.50 in · $15.00 out / 1M tokens

    $0.0132
  22. Claude Opus 5

    Anthropic · $5.00 in · $25.00 out / 1M tokens

    $0.0260
  23. GPT-5.6 Sol

    OpenAI · $5.00 in · $30.00 out / 1M tokens

    $0.0264
  24. GPT-5.5

    OpenAI · $5.00 in · $30.00 out / 1M tokens

    $0.0264

Speech-to-text

$0.0020to$0.0240/ min
10 options

What it hears. The spread here is mostly accents, background noise and how fast a word is recognised — on a clean line the cheapest is hard to fault.

  1. Soniox RT v5

    Soniox · $0.0020 / min

    $0.0020
  2. Ink Whisper

    Cartesia · $0.0022 / min

    $0.0022
  3. Grok STT

    xAI · $0.0033 / min

    $0.0033
  4. Scribe v2 Realtime

    ElevenLabs · $0.0042 / min

    $0.0042
  5. Nova 2

    Deepgram · $0.0058 / min

    $0.0058
  6. Nova 2 Conversational AI

    Deepgram · $0.0058 / min

    $0.0058
  7. Nova 2 Phone Call

    Deepgram · $0.0058 / min

    $0.0058
  8. Nova 3 Medical

    Deepgram · $0.0077 / min

    $0.0077
  9. Nova 3

    Deepgram · $0.0092 / min

    $0.0092
  10. Amazon Transcribe

    AWS · $0.0240 / min

    $0.0240

Text-to-speech

$0.0105to$0.0350/ min
11 options

What it sounds like. The one line where paying more is most audible to a caller, and the one worth trying before you decide on price.

  1. Grok TTS

    xAI · $0.0150 / 1K chars

    $0.0105
  2. Gemini 2.5 Flash TTS

    Google · $0.0170 / 1K chars

    $0.0119
  3. Rime Mist v2

    Rime · $0.0300 / 1K chars

    $0.0210
  4. Gemini 2.5 Pro TTS

    Google · $0.0340 / 1K chars

    $0.0238
  5. Sonic Turbo

    Cartesia · $0.0374 / 1K chars

    $0.0262
  6. Sonic 2

    Cartesia · $0.0374 / 1K chars

    $0.0262
  7. Sonic 3

    Cartesia · $0.0374 / 1K chars

    $0.0262
  8. Rime Arcana

    Rime · $0.0400 / 1K chars

    $0.0280
  9. Flash v2.5

    ElevenLabs · $0.0500 / 1K chars

    $0.0350
  10. Conversational v3

    ElevenLabs · $0.0500 / 1K chars

    $0.0350
  11. Turbo v2.5

    ElevenLabs · $0.0500 / 1K chars

    $0.0350

Telephony

$0.0000to$0.0140/ min
3 options

The carrier leg. Outbound is what the estimate uses; the inbound rate for each is noted beneath it.

  1. WhatsApp outbound

    WhatsApp · $0.0000 inbound

    $0.0000
  2. SIP outbound

    SIP · $0.0040 inbound

    $0.0040
  3. Twilio outbound

    Twilio · $0.0085 inbound

    $0.0140
Phone numbers
$1.15
Monthly cost varies by country and number type. Price shown is for US local numbers.
Recording storage
$0.5000
Billed based on actual storage duration. Minimum 1 day charged.
Get started

Start on credit. Add money when it runs out.

No card, no call with us. Build an agent, talk to it in the browser, and only pay once real calls start going out.