Millis AI Pricing Explained (2026): What a Minute Costs
Millis AI has no pricing page on its main site (millis.ai/pricing returns not found), but its docs publish the full breakdown: a $0.02-a-minute platform fee, plus the language model billed by tokens, the voice billed by the characters your agent speaks, and speech-to-text at $0.0043 a minute. Your phone carrier bills the line. Every price below is from docs.millis.ai/pricing and millis.ai, read October 25, 2026; the measured numbers are from Cozzo's own billed test calls.
On Millis, the $0.02 platform fee is the smallest part of most bills. Pick your voice first: an ElevenLabs voice adds about 5¢ a minute, while an OpenAI or Deepgram voice adds under 1¢.
Millis AI charges $0.02 a minute plus the LLM, voice and speech-to-text you pick. Every line of a Millis minute, monthly costs, and what our test calls cost.
What you pay for on every Millis AI minute
Millis platform fee
$0.02/min
Millis's base charge for every minute on its platform, on top of the model, voice and speech-to-text lines below. Its homepage advertises voice AI "starting at just $0.02/min".
Language model (LLM)
$0–$0.008/min
Billed by tokens. Millis's docs estimate about $0.004 a minute for GPT-4o, $0.008 for GPT-4 Turbo, $0.0004 for GPT-3.5 Turbo and $0.00018 for Llama 3. There's no Millis charge if you connect your own LLM; your model provider bills you instead.
Voice (text-to-speech)
~$0.0005–$0.125/min
Billed by the characters your agent speaks. Per 1,000 characters: ElevenLabs $0.10 (about $0.05/min), Rime $0.075 (~$0.0375), Cartesia $0.0392 (~$0.0196), OpenAI and Deepgram $0.015 (~$0.0075), PlayHT $0.25 (~$0.125) and Neets $0.001 (~$0.0005). Millis's per-minute figures assume about 500 characters a minute.
Speech-to-text
$0.0043/min
$0.0043 a minute for turning the caller's speech into text. In Millis's own worked example it's charged only for the minutes the caller is talking.
Phone number and carrier minutes
Your carrier's rate
You import a Twilio, Vonage, Plivo or Exotel number, or route calls to Millis over SIP, and that carrier bills its own charges. Millis's API can also buy numbers, but it doesn't publish a price for them.
Total per minute
about 2.5¢ to 15.7¢
Millis's own estimates per minute of call. The voice you pick is the biggest swing; your carrier's charges come on top.
What a month costs: Millis's numbers and our measured rate
500 min/month
- OpenAI mix
- $18
- ElevenLabs mix
- $39
- Cozzo measured
- $45
1,000 min/month
- OpenAI mix
- $36
- ElevenLabs mix
- $78
- Cozzo measured
- $90
2,500 min/month
- OpenAI mix
- $90
- ElevenLabs mix
- $196
- Cozzo measured
- $225
5,000 min/month
- OpenAI mix
- $179
- ElevenLabs mix
- $392
- Cozzo measured
- $450
10,000 min/month
- OpenAI mix
- $358
- ElevenLabs mix
- $783
- Cozzo measured
- $900
The two mix columns are Cozzo's arithmetic from Millis's own per-minute estimates, both with GPT-4o: $0.02 platform + $0.004 GPT-4o + $0.0075 OpenAI voice or $0.05 ElevenLabs voice + $0.0043 speech-to-text, so $0.0358 or $0.0783 a minute at Millis's 500 characters a minute. “Cozzo measured” uses the ~$0.09 a minute Millis actually billed on Cozzo's Oct 20 test calls (its default LLM with an ElevenLabs voice). None include your carrier's number and minutes or a volume discount. Want your own volume and carrier? Try the cost calculator.
Packages and add-ons
Pay as you go
$0.02/min + usage
The only pricing Millis publishes: the platform fee plus the LLM, voice and speech-to-text you use. Its site and docs list no monthly plans, minimums or contracts.
Millis's own example
$0.66 per 10 minutes
Millis's docs work through a 10-minute call with GPT-4o and an ElevenLabs voice: $0.20 platform + $0.04 LLM + $0.40 voice (4,000 characters) + $0.0215 speech-to-text = $0.6615, about 6.6¢ a minute. The voice is 60% of that bill.
Bring your own LLM
No Millis LLM charge
Connect your own model over a WebSocket and Millis doesn't bill the LLM line; you pay your provider. "Choose by Millis" instead picks the lowest-latency model from its list for you.
ElevenLabs volume discount
Up to 25% off
Millis's docs say high-usage customers can get ElevenLabs voice costs cut by up to 25%. You have to ask Millis for it.
Phone numbers
Your carrier's rate
Import a Twilio, Vonage, Plivo or Exotel number in the Millis dashboard, or connect your phone system over SIP. Calls can run in Millis's US-West or EU-West region.
Not published
Ask Millis
Free trial minutes or credits, a limit on calls at once, how partial minutes are rounded, an enterprise plan, uptime terms, and SOC 2, HIPAA or BAA status.
What does one call cost? Voice AI cost per call, measured on our real phone calls →
How it compares on published pricing
- MMillis AI
- Starting price
- From $0.02/min
- Best for
- Builders who want phone agents with no-code or code
Retell AI- Starting price
- $0 platform fee; $0.07–$0.31/min for voice agents
- Best for
- Teams that want a no-code flow builder plus an API with itemized per-minute pricing
Vapi- Starting price
- $0 to start; $0.05/min Vapi hosting, models billed pass-through at cost
- Best for
- Developers who want to choose their own LLM, transcriber, voice and telephony provider
Ultravox- Starting price
- $0.05/min (first 30 min free)
- Best for
- Developers wanting a speech-native model at a flat per-minute rate
Bland AI- Starting price
- $0.14/min, $0 platform fee
- Best for
- Teams that want one flat per-minute price with no model pass-through bills
ElevenLabs Agents- Starting price
- Agents: $0.08 per minute
- Best for
- Builders who want ElevenLabs' voices and speech models in one agent pipeline
Pricing explained, line by line:Retell AI pricing →Vapi pricing →Ultravox pricing →Bland AI pricing →ElevenLabs Agents pricing →
Cozzo tested: reply time on real phone calls
Median time from the end of the caller’s speech to the agent’s first reply, on each vendor’s default model and voice (Oct 20, 2026). Places are out of 8 platforms tested.
- Retell AI: 1.04 s (tied 1 of 8, 11 calls)
- ElevenLabs Agents: 1.04 s (tied 1 of 8, 11 calls)
- Ultravox: 1.18 s (3 of 8, 11 calls)
- Millis AI: 1.25 s (4 of 8, 12 calls)
- Vapi: 1.56 s (7 of 8, 11 calls)
- Bland AI: 1.86 s (8 of 8, 11 calls)
See the full phone-call benchmark → Small samples, so treat close numbers as ties.
What to check before you choose
- Pick the voice first. On Millis the voice is usually the biggest line: an ElevenLabs voice is about $0.05 a minute against about $0.0075 for an OpenAI or Deepgram voice, so it can cost more than double the $0.02 platform fee.
- Chatty agents cost more. Millis bills the voice by characters spoken, so long answers raise the bill. Its per-minute figures assume about 500 characters a minute, while its own 10-minute example assumes about 800 a minute of agent speech (4,000 characters in 5 minutes).
- Treat the LLM estimates as rough. Millis's docs price older models (GPT-4o, GPT-4 Turbo, GPT-3.5 Turbo, Llama 3). Check the dashboard for the model you actually use, or bring your own LLM and pay your provider directly.
- Budget for the phone line. Millis's rates don't include a number: your Twilio, Vonage, Plivo or Exotel account, or your SIP provider, bills the number and the call minutes.
- Ask before you scale or handle patient data. Millis doesn't publish a limit on calls at once, uptime terms or a BAA. Retell includes 20 calls at once and a BAA you can self-sign; Vapi sells HIPAA as a $2,000 a month add-on.
- Compare whole minutes. Retell lists $0.07 to $0.31 a minute, Vapi $0.05 a minute for hosting plus the models you pick, Ultravox a flat $0.05 with its model and voices included, Bland AI $0.14 on its Start plan and ElevenLabs Agents $0.08 plus the LLM. On Cozzo's Oct 20 phone calls Millis billed about $0.055 per ~35-second call (about $0.09 a minute) and replied in a median 1.25 seconds over 12 calls.
Questions people ask
How much does Millis AI cost per minute?
$0.02 a minute for the platform, plus the LLM, voice and speech-to-text. On Millis's own estimates that comes to about 2.5¢ to 15.7¢ a minute, mostly depending on the voice: about 3.6¢ with GPT-4o and an OpenAI voice, or 7.8¢ with an ElevenLabs voice. On Cozzo's test calls with its default setup, Millis billed about 9¢ a minute. Your carrier's charges come on top.
Does Millis AI have a pricing page?
Not on its main site: millis.ai/pricing returns a not-found page (checked October 25, 2026). The pricing is in its docs at docs.millis.ai/pricing, and the homepage says "starting at just $0.02/min".
Is Millis AI free to try?
Millis doesn't publish free minutes or trial credits on its site or in its docs. Check the dashboard when you sign up, or ask Millis.
Why does the voice cost more than the platform on Millis?
Because Millis passes the voice through by characters spoken. ElevenLabs is $0.10 per 1,000 characters, about $0.05 a minute at Millis's 500-characters-a-minute estimate, which is more than double the $0.02 platform fee. OpenAI and Deepgram voices are $0.015 per 1,000 characters, about $0.0075 a minute.
Does Millis AI charge for phone numbers?
Millis doesn't publish a phone-number price. You can import a Twilio, Vonage, Plivo or Exotel number, or route calls over SIP, and your carrier bills the number and minutes. Millis's API can buy numbers, but the price isn't listed.
Is Millis AI cheaper than Retell or Vapi?
On Cozzo's test calls with each vendor's default setup, Millis billed about $0.09 a minute, Vapi about $0.08–$0.09 and Retell about $0.15. On list prices it depends on your picks: Millis is $0.02 a minute plus models, Vapi $0.05 a minute hosting plus models, and Retell $0.07 to $0.31 a minute.
How fast is Millis AI?
Millis advertises 500 to 600 ms latency (both figures appear on its homepage). On Cozzo's Oct 20 phone calls, measured end to end over a real phone line, its median reply was 1.25 seconds over 12 calls. In our Round 4 receptionist calls it replied in a median 1.31 seconds and passed 12 of 15 calls.
Does Millis AI sign a BAA for HIPAA?
Millis doesn't publish HIPAA, SOC 2 or BAA terms on its site or in its docs, so ask Millis directly before you send it patient data.
Ready to look closer? See Millis AI’s full listing and price history →
Not sure which fits? Answer a few quick questions in the finder →
Prices from vendors' official pricing pages, checked October 25, 2026. Not sponsored; order isn't a ranking.