Audio in
$3per 1M
$0.005/min
Text in $0.50
to $0.75
Audio out
$12per 1M
$0.018/min
Text out $2
to $4.50
Blended / min
1.3cest
40% listen, 60% speak
Estimate, range 1 to 2c
before telephony
Free tier
$0tokens
Rate limited
For prototyping
paid above
Per minute figures are Google's own equivalents for audio tokens. Telephony is not included, and the API is a model, not a phone agent framework.
Gemini Live API pricing by token: audio, text, and grounding
The Gemini Live API is a speech to speech model billed per token, with per minute equivalents Google publishes. Google prices each part as follows.
The audio the model listens to is billed at about $3 per million tokens, roughly $0.005 a minute, so the cost of a session rises with how much the caller talks, not just how long the call lasts.
The audio the model speaks is the larger line at about $12 per million tokens, roughly $0.018 a minute, so a chatty agent that talks more than it listens costs noticeably more than the input rate suggests.
Text tokens cost far less than audio. Input runs $0.50 to $0.75 and output $2 to $4.50 per million tokens, so mixing text and audio can cut the effective per-minute cost.
Live API tokens are free on the free tier within rate limits, enough to prototype an agent, but the limits make it unsuitable for steady production traffic, where usage moves onto the paid token rates above.
Google Search grounding requests, shared across the newer models, are free up to 5,000 then $14 per thousand, an extra line that adds up for agents that look things up on most turns.
Rates checked on Google's Gemini API pricing page on 22 October 2026. A blended voice minute depends on how much the model listens and speaks, so per minute figures are a guide.
How much the Gemini Live API really costs per minute
Because it bills per token, the cost of a call depends on speech length and how talkative the agent is. A Gemini Live minute is built from these pieces.
Which Gemini Live plan fits your use case
Free tokens within rate limits let developers build and test a live voice agent before paying, which suits a proof of concept but not steady production traffic, where the rate limits and token metering start to bite.
The paid native audio tier works out to an estimated $0.013 a minute at a 40% listening and 60% speaking split, inside a $0.01 to $0.02 range. Audio input is $0.005 a minute and output is $0.018, so a chatty agent sits at the upper end of that range.
Text tokens cost far less than audio, at $0.50 to $0.75 per million input and $2 to $4.50 per million output. A session that handles routine steps as text and speaks only when a caller needs it costs less, which rewards designs that keep voice for the moments that need it.
Google does not include telephony in the Live API. Your team connects a carrier or SIP provider, which bills per minute on top of the token cost, and that extra layer belongs in any true cost comparison.
Volume pricing and custom terms come through Google Cloud and Vertex AI. Teams that want the Live API inside a broader Google Cloud commitment use that route instead of the pay-as-you-go developer tier, and they should ask Google for the rate that applies to them.
Gemini Live API pricing calculator: estimate your monthly bill
Gemini figure assumes a call where the model listens 40% and speaks 60% of the time: 0.4 x $0.005 plus 0.6 x $0.018 gives about $0.013 a minute, before telephony. PyAI figure is minutes times the per minute rate, with telephony on the same bill.
Gemini Live hidden costs: tokens, telephony, and grounding
- Token pricing varies: a call's cost depends on how long the model listens and speaks, so a minute is an estimate, not a fixed price.
- Output costs more: audio output at $12 per million is four times the input, so a talkative agent costs more.
- Telephony is separate: the API is a model, so you add a carrier or SIP provider and pay their per minute rates on top.
- You build the orchestration: there is no phone agent framework, so call handling, barge in, and transfers are yours to build.
- Search grounding bills: grounding is free for 5,000 requests a month, then $14 per 1,000, if you enable it.
- Free tier limits: the free tokens are rate limited, so production workloads move to the paid tier.
- Thinking tokens count: output pricing includes thinking tokens, which add to the bill on reasoning heavy turns.
Gemini Live API reviews: what users say
We looked for real, attributed user reviews of Gemini Live API that mention pricing and the product. As of October 2026 we found no public user reviews on the sources below, so there is nothing to quote honestly. The Gemini Live API is a developer feature of Google's Gemini models rather than a separately reviewed product, so the Gemini reviews on G2 are about the assistant and its models, not the real-time voice API. The Live API has no public review base of its own to quote from.
No public user reviews found
Evaluate Gemini Live API on a short trial and a reference call rather than on ratings, since no verifiable user reviews exist yet.
Gemini Live API vs PyAI: price and billing compared
| Google Gemini Live | PyAI | |
|---|---|---|
| Billing model | Per token | ✓ Per minute, a fixed price |
| Voice rate | About $0.013/min blended, est, the lower rate | Omni $0.05/min + $0.01 telephony, Agents $0.08/min, fixed |
| Predictability | Varies with tokens | ✓ A minute is a minute |
| Free tier | ✓ Free tokens, rate limited | $50 credit, no credit card |
| Telephony | Not included | ✓ Telephony per minute, on the same bill |
| Phone agent layer | You build it | ✓ Built in phone agent |
| Speech to text and TTS | Within the model | ✓ PyAI Hear at $0.001 and Speak at $0.04, itemized per minute |
| Model quality | ✓ Google frontier models | Hosted voice, OpenAI compatible |
| Output formats | Audio streams | ✓ PyAI encodes WAV, PCM or G.711 for the phone network |
| Reviews | No review base for the API itself | No G2 profile yet |
Choose Gemini Live if you want
- A low per minute rate for a speech to speech model.
- A free tier for prototyping and Google models.
- Full control to build your own voice stack.
Choose PyAI if you want
- A fixed per minute price instead of token billing.
- Telephony and a phone agent on the same bill.
- WAV, PCM or G.711 output, where Gemini Live returns raw audio streams.
Gemini Live cost by volume: voice minutes a month
Since the Live API bills per token, a monthly estimate depends on how talkative the agent is. The Gemini figures use a blended per minute estimate.
| Use case | What you need | Tier | Watch out for |
|---|---|---|---|
| Prototyping | Testing at low volume | Free tier | Rate limits cap steady production traffic. |
| Cost sensitive | High volume voice | Paid native audio | Telephony is billed as a separate line. |
| Phone calls | Telephony and orchestration | Paid audio plus your own telephony | Carrier minutes are billed on top of the token cost. |
Gemini Live vs PyAI monthly cost by volume
Gemini uses a blended per minute estimate of about $0.013, inside its $0.01 to $0.02 range. PyAI Omni is minutes times the all-in rate of $0.05/min + $0.01 telephony. Gemini Live's estimated rate is lower per minute, as the table shows, while PyAI is a fixed, predictable price with telephony on the same bill.
| Voice minutes a month | Gemini Live (est) | PyAI Omni | PyAI Agents |
|---|---|---|---|
| 1,000 | $13 | $60 | $80 |
| 10,000 | $130 | $600 | $800 |
| 50,000 | $650 | $3,000 | $4,000 |
| 100,000 | $1,300 | $6,000 | $8,000 |
The Gemini figures use the same 40% listening, 60% speaking split, about $0.013 per voice minute, with telephony left out. PyAI Omni is $0.05/min + $0.01 telephony, with telephony on the same bill.
Gemini Live API pricing: the bottom line
The Gemini Live API is priced per token, with audio input at $3 per million, about $0.005 a minute, and output at $12 per million, about $0.018, so a blended voice minute is roughly $0.01 to $0.02, with a free tier for prototyping. Telephony and orchestration are yours to add. PyAI bills per minute instead, with Omni at $0.05/min + $0.01 telephony and Agents at $0.08/min, a fixed price with telephony on the same bill. Sandbox keys are free, with $50 in credit and no credit card.
PyAI does not match every Gemini Live API feature, such as Google's frontier models, Search grounding, and a free token tier. PyAI instead bills voice infrastructure by the minute, putting speech to text, text to speech, and a phone agent on one invoice.
Gemini Live API pricing FAQs
The questions teams ask most before choosing the Gemini Live API.
How much does the Gemini Live API cost?
The Gemini Live API is priced per token, and Google also lists per minute equivalents. Audio input is $3 per million tokens, about $0.005 a minute, and audio output is $12 per million, about $0.018 a minute. Text input is $0.50 to $0.75 per million. A blended voice minute works out to roughly $0.01 to $0.02.
Does the Gemini Live API have a free tier?
Yes. Google offers a free tier where Live API input and output tokens are free of charge, subject to rate limits, which is useful for prototyping. The paid tier applies above those limits. Google Search grounding gives 5,000 free requests a month across newer models, then $14 per 1,000 requests.
How much is a Gemini Live voice minute?
Google lists audio input at about $0.005 a minute and audio output at about $0.018 a minute, so a blended voice minute, where the model listens and speaks, is roughly $0.01 to $0.02 depending on how talkative the agent is. For a speech to speech model, that is a low per minute rate before any telephony is added.
Does the Gemini Live API include telephony?
No. The Gemini Live API is a model API, so telephony is not included. You connect your own carrier or SIP provider and pay their per minute rates on top. The API also does not bundle a phone agent framework, so call handling, barge in, and transfers are yours to build around the model.
What is the difference between Gemini Live and the Gemini app?
The Gemini app is a consumer assistant you talk to, while the Gemini Live API is the developer product for building voice into your own software, billed per token. The pricing on this page covers the API. For a fixed per minute price with telephony included, a voice infrastructure provider like PyAI is the closer fit.
What is a more predictable alternative to the Gemini Live API?
PyAI bills per minute instead of per token, so a minute is a fixed price. Omni speech to speech is $0.05/min + $0.01 telephony and Agents is $0.08/min, with telephony as its own line on the same bill and a predictable invoice. PyAI is easier to forecast than token billing, and the phone agent layer comes built in.
More Gemini voice comparisons and guides
Price PyAI against your own volume.
Per minute pricing with no plan fee and no credits to burn. Speak is $0.04/min, Cast $0.02/min, Hear $0.001/min, and Omni $0.05/min + $0.01 telephony, with $50 in starter credit and no credit card.
