Speech-to-speech is here: Learn about Prism →
pricing

Conversation-grade voice models for customer-obsessed brands.

Designed to keep customers on the line.

Free

Try every Rime model on real calls

$0/mo

No card required

Get started for free
  • $50 of usage every month, renews monthly
  • Prism, Mist v3, Coda v2 and Rime STT
  • Move to Flex when you need more

Flex

Billed by minute, monthly to your card

From$0.002/min

No monthly fee, no commitment

Start with Flex
  • PrismSpeech to speech$0.080 / min
  • Rime STTSpeech to text$0.002 / minFree until Nov 19
  • Mist v3Text to speech$0.030 / min
  • Coda v2Text to speech$0.050 / min

Scale

Steady usage, 20% off every rate

From$0.0016/min

$100 a month

Start with Scale
  • PrismSpeech to speech$0.064 / min
  • Rime STTSpeech to text$0.0016 / minFree until Nov 19
  • Mist v3Text to speech$0.024 / min
  • Coda v2Text to speech$0.040 / min
  • $100 of credit every month, rolls over up to $200
  • Usage past your credit billed at the same rates

Enterprise

High volume, hands-on support

Custom

Volume pricing, billed annually

Schedule a call
  • Volume rates on every model
  • Cloud, VPC, or on-prem deployment
  • SLAs with a Forward Deployed Engineer and Linguist
  • BAA (HIPAA) and SOC 2 Type II report
  • Custom voices

Compare plans by model

Free
Flex
Scale
Enterprise
Pricing
PrismSpeech to speech, metered by the minute of call audio
Included in your $50
$0.080 / min
$0.064 / min
Volume rates
Included usage
$50 a month, renews monthly
None, pay for what you use
$100 of credit a month, rolls over up to $200
Annual commitment
Billing
No card required
Metered by the minute, billed monthly to your card
$100 a month to your card, overage at the same rates
Invoiced
Scale
DeploymentWhere Prism runs
Rime cloud
Rime cloud
Rime cloud
Cloud, VPC, or on-prem
StreamingReal-time audio in and out over WebSockets, on every plan
✓
✓
✓
✓
Framework integrationsLiveKit, Pipecat, and other agent platforms, on every plan
✓
✓
✓
✓
Fine-tuning on your callsTrained on your transcripts and tool-call logs, measured against your outcomes
—
—
—
✓
Custom voicesPrism speaks in your voice
—
—
—
✓
Support and compliance
Support
Docs and community
Email
Email
SLA, Forward Deployed Engineer and Linguist
HIPAABAA
—
—
—
✓
SOC 2 Type IIReport on request
—
—
—
✓
Model capabilities
Reasons over audioUnderstands the caller's tone as well as their words
✓
✓
✓
✓
Text before speechEvery response is generated as text first, so your guardrails and disclosures can check it before the caller hears it
✓
✓
✓
✓
Tool callsLook things up and take action mid-call
✓
✓
✓
✓
Turn takingHalf duplex at launch, with interruption handling
✓
✓
✓
✓
Telephony-ready8 kHz phone audio in and out
✓
✓
✓
✓

Benchmarks

How Prism compares, on every plan

The same model at every tier. On Artificial Analysis's speech-to-speech leaderboard, Prism starts speaking in 0.57 seconds and scores 56.6 on τ-Voice, ahead of every realtime model from OpenAI and Google that isn't reasoning first.

See the full breakdown on the Prism page
0.4 s 0.6 s 0.8 s 1.0 s 1.2 s 1.4 s 20% 30% 40% 50% 60% 70% Time to first audio · ◀ faster τ-Voice task completion · better ▶ Faster and more accurate Prism56.6% · 0.57 s GPT-Live-1 (Sol, low)59.3% · 1.24 s · reasoning on GPT-Realtime-2.1 High45.7% · 1.21 s · reasoning on Gemini 3.8 Live30.1% · 1.18 s · reasoning off Reasoning offReasoning on 0.4 s 0.6 s 0.8 s 1.0 s 1.2 s 1.4 s 20% 30% 40% 50% 60% 70% Time to first audio · ◀ faster τ-Voice task completion · better ▶ Faster and more accurate Prism56.6% · 0.57 s GPT-Live-1 (Sol, low)59.3% · 1.24 s GPT-Realtime-2.1 High45.7% · 1.21 s Gemini 3.8 Live30.1% · 1.18 s Reasoning offReasoning on

Each dot is a speech-to-speech model on Artificial Analysis's τ-Voice benchmark. Up is more accurate, left is faster to start speaking, and rings mark models that had reasoning turned on. Prism gets to the top-left corner without it.

Free
Flex
Scale
Enterprise
Pricing
Mist v3Text to speech, lowest latency. About 1K characters per minute.
Included in your $50
$0.030 / min
$0.024 / min
Volume rates
Coda v2Text to speech, most natural. About 1K characters per minute.
Included in your $50
$0.050 / min
$0.040 / min
Volume rates
Included usage
$50 a month, renews monthly
None, pay for what you use
$100 of credit a month, rolls over up to $200
Annual commitment
Billing
No card required
Metered by the minute, billed monthly to your card
$100 a month to your card, overage at the same rates
Invoiced
Scale
DeploymentSelf-host via Docker Compose or Kubernetes
Rime cloud
Rime cloud
Rime cloud
Cloud, VPC, or on-prem
Custom voice cloning
—
—
—
✓
Support and compliance
Support
Docs and community
Email
Email
SLA, Forward Deployed Engineer and Linguist
HIPAABAA
—
—
—
✓
SOC 2 Type IIReport on request
—
—
—
✓
Model capabilities
LanguagesProduction-ready languages
Coda v2English, Spanish, Portuguese, French, German, Italian, Dutch, Japanese, Arabic, HindiMist v3English, French, German, Spanish
Number of voicesCurated vocal roster
Coda v2313Mist v378
Spell functionalityExplicit character-by-character spelling
✓
✓
✓
✓
Speed adjustmentContinuous playback rate modifiers
✓
✓
✓
✓
Pronunciation controlNames, addresses, and alphanumerics, rendered right
Coda v2Custom lexicon with typed or recorded phonemesMist v3Inline pronunciation control for English only
Custom pauseGranular control of phrasing intervals
Coda v2Via punctuationMist v3✓
TTFA (P50)Time to first audio, one request at a time, measured at the server without the network round trip
Coda v230 msMist v337 ms
TTFA (P90)Time to first audio, one request at a time, measured at the server without the network round trip
Coda v2—Mist v356 ms
TTFA (P95)Time to first audio, one request at a time, measured at the server without the network round trip
Coda v260 msMist v3—
StreamingStream audio as it's generated
HTTP, WebSockets, and gRPC
Word-level timestampsPrecise interruption handling for voice agents
Coda v2WebSockets and gRPCMist v3WebSockets and gRPC, English
Framework integrationsDrop into your existing voice-agent stack
LiveKit, Pipecat, and many other agent platforms
Free
Flex
Scale
Enterprise
Pricing
Rime STTSpeech to text, metered by audio duration. Free on every plan until Nov 19.
Included in your $50
$0.002 / min
$0.0016 / min
Volume rates
Included usage
$50 a month, renews monthly
None, pay for what you use
$100 of credit a month, rolls over up to $200
Annual commitment
Billing
No card required
Metered by the minute, billed monthly to your card
$100 a month to your card, overage at the same rates
Invoiced
Scale
DeploymentSame deployment as Rime TTS
Rime cloud
Rime cloud
Rime cloud
Cloud, VPC, or on-prem
Support and compliance
Support
Docs and community
Email
Email
SLA, Forward Deployed Engineer and Linguist
HIPAABAA
—
—
—
✓
SOC 2 Type IIReport on request
—
—
—
✓
Model capabilities
LanguagesSet per request
English, Spanish
Interim resultsPartial transcripts while the speaker is still talking
Rewrite as it goes, or add as it goes (live only)
FormattingTwo output modes
Clean (digits, fillers removed) or word for word
Custom vocabularyProduct names, jargon, proper nouns
Up to 1,024 characters per request, no extra charge
Turn detectionDeciding when the speaker is done
Handled by your orchestration layer
Time to final transcriptFrom end of speech, P50 / P90, on one NVIDIA L4 with no network in the path
Live105 ms / 107 msFile206 ms / 207 ms for a 5 s file
Word error rateFLEURS, Open ASR Leaderboard method
Live5.1% English, 6.2% SpanishFile4.5% English, 5.3% Spanish
ThroughputPer NVIDIA L4
About 810 seconds of audio per second, roughly 24x real time
ProtocolJSON or protobuf back
LiveWebSocket or gRPC streamingFilegRPC
Audio inWhat you send
Live16 kHz mono PCMFileWAV up to 16 MB (about 8 minutes)
Session limitsPer connection or call
LiveOne utterance per connection, up to 2 minutes, 15 s silence timeoutFileOne file per call
SDKsQuickstarts in the docs
Python and Node

Not sure which model fits?

Talk to a voice specialist about your latency budget, languages, and deployment. We'll help you pick in one call.

Schedule a callTry for free

Questions, answered

How is Rime priced?
Every account starts with $50 of usage free every month, across every Rime model, with no card required. Past that, the Flex plan is pay as you go by the minute of audio, with no minimum: Mist v3 is $0.03 per minute, Coda v2 is $0.05 per minute, Prism is $0.08 per minute, and Rime STT is free until November 19, then $0.002 per minute. Text to speech is metered by characters generated, about 1,000 characters per minute. The Scale plan takes 20% off every rate for $100 a month, and Enterprise plans move to volume pricing on an annual commitment.
Can I get volume discounts for Rime?
Yes. The Enterprise plan is built around volume pricing for organizations running at scale, with SLAs and dedicated support included. Talk to us and we'll shape a plan around your usage.
Is speech to text really free?
Yes. Rime STT is free on every plan through November 19, 2026. From November 20 it is $0.002 per minute on Flex and $0.0016 on Scale, metered by audio duration. Concurrency limits still apply during the free period.
What happens when I use up the free $50?
Your free usage renews each month and doesn't roll over. To keep going before then, move to Flex: add a card and you're billed pay as you go at the rates above, with no monthly fee and nothing to cancel. If your usage is steady, Scale gives you $100 of credit a month at 20% off every rate, with unused credit rolling over up to $200.
Does Rime have an open API?
Yes. Text to speech streams over HTTP, WebSockets, and gRPC, speech to text over WebSockets and gRPC, and Prism over WebSockets. All of them can self-host on your own hardware when compliance demands it. Most teams are up and running the same afternoon.

Start the conversation

Try for freeBook a call
product
Book a demoPricingResourcesDocs
Company
AboutCareersPrivacyTermsSecuritySubprocessors
account
Sign upLog in

Guides

Prism vs OpenAIElevenLabs vs RimeRime vs DeepgramEnterprise TTS buyer's guideOn-prem voice AI guideTTS voice best practices
Social
LinkedinXYouTube
HIPAACOMPLIANTSOC 2COMPLIANT
© 2026 Rime Labs