Skip to content
Model infoModelsSnapshot: 15 Oct 2026

Cartesia / Ink

Cartesia Ink 2

Speech recognition through Cartesia’s Ink service.

Transcription
availableClosed weightsSpeech-to-text
Official documentation (opens in a new tab)Reviewed 12 Oct 2026
10 of 20 specification fields collected
Output formatsJSON transcriptSee specification notes below
Languages5 listedFull language list in specifications
Input billingSee ratesCartesia
Output billingSee ratesPublished units and conditions below

Specifications

Provider-documented facts. “Not collected” means we have not verified that field yet, not that the feature is unsupported.

Context / input limit
Not collected
Maximum output
Not collected
Input modalities
AudioSource for Input modalities (opens in a new tab)
Output modalities
TextSource for Output modalities (opens in a new tab)
Reasoning
Not collected
Tool calling
Not collected
Structured output
Not collected

JSON transcripts and turn events do not establish configurable schema-constrained generation.

Streaming
SupportedSource for Streaming (opens in a new tab)
Batch processing
Not supportedSource for Batch processing (opens in a new tab)

Ink 2/Preview support streaming; the file transcription endpoint is restricted to Ink Whisper.

Fine-tuning
Not collected
Knowledge cutoff
Not collected
Release date
2026-09-17Source for Release date (opens in a new tab)

Date in the current model table. An earlier public Ink 2 launch article is dated July 9, 2026; this is not asserted as the first-ever family launch.

Parameters
Not collected
Architecture
Not collected
Languages
en, fr, hi, ja, esSource for Languages (opens in a new tab)
Context caching
Not collected
Output formats
JSON transcriptSource for Output formats (opens in a new tab)
Automatic turn detection
SupportedSource for Automatic turn detection (opens in a new tab)
Custom vocabulary
Up to 100 keyterms; 1,200 characters combinedSource for Custom vocabulary (opens in a new tab)
Usage metering
3 credits per second of audioSource for Usage metering (opens in a new tab)

Includes silence.

Access & pricing

Prices belong to an endpoint and billing tier. A provider listing does not guarantee access for every account or region.

Cartesia

Provider source (opens in a new tab)
API model ID
ink-2
Access
Native API with an eligible account and API key or access token. Credit balance and plan limits apply. No paid inference was performed.
Regions
api.cartesia.ai routes to the closest endpoint. Dedicated US, EU, UK, India and Australia deployments require Enterprise access.
API endpoint URL
wss://api.cartesia.ai/stt/websocketSource for API endpoint URL (opens in a new tab)
Usage metering
3 credits per second of audioSource for Usage metering (opens in a new tab)

Silence is billable.

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 12 Oct 2026
CategoryUsageRateUnit
Subscription credit overageScale overage1$38.001M credits
Subscription credit overageStartup overage2$45.001M credits
Subscription credit overagePro overage3$65.001M credits

Rate conditions

  1. Scale subscription required. Charged only beyond included credits with overages enabled; monthly subscription fees are additional. These are credit prices, not per-million-token or per-million-character rates. Convert using this endpoint's documented credit meter.
  2. Startup subscription required. Charged only beyond included credits with overages enabled; monthly subscription fees are additional. These are credit prices, not per-million-token or per-million-character rates. Convert using this endpoint's documented credit meter.
  3. Pro subscription required. Charged only beyond included credits with overages enabled; monthly subscription fees are additional. These are credit prices, not per-million-token or per-million-character rates. Convert using this endpoint's documented credit meter.
Pricing source (opens in a new tab)

Data & enterprise terms

Enterprise customers may opt into ZDR for TTS/STT inference content; operational metadata is retained. Voice cloning and fine-tuning are excluded. General terms permit model improvement/training unless separately agreed or an opt-out is processed.

Read the service policy (opens in a new tab)

Cartesia

Provider source (opens in a new tab)
API model ID
ink-2
Access
Native API with an eligible account and API key or access token. Credit balance and plan limits apply. No paid inference was performed.
Regions
api.cartesia.ai routes to the closest endpoint. Dedicated US, EU, UK, India and Australia deployments require Enterprise access.
API endpoint URL
wss://api.cartesia.ai/stt/turns/websocketSource for API endpoint URL (opens in a new tab)
Usage metering
3 credits per second of audioSource for Usage metering (opens in a new tab)

Silence is billable.

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 12 Oct 2026
CategoryUsageRateUnit
Subscription credit overageScale overage1$38.001M credits
Subscription credit overageStartup overage2$45.001M credits
Subscription credit overagePro overage3$65.001M credits

Rate conditions

  1. Scale subscription required. Charged only beyond included credits with overages enabled; monthly subscription fees are additional. These are credit prices, not per-million-token or per-million-character rates. Convert using this endpoint's documented credit meter.
  2. Startup subscription required. Charged only beyond included credits with overages enabled; monthly subscription fees are additional. These are credit prices, not per-million-token or per-million-character rates. Convert using this endpoint's documented credit meter.
  3. Pro subscription required. Charged only beyond included credits with overages enabled; monthly subscription fees are additional. These are credit prices, not per-million-token or per-million-character rates. Convert using this endpoint's documented credit meter.
Pricing source (opens in a new tab)

Data & enterprise terms

Enterprise customers may opt into ZDR for TTS/STT inference content; operational metadata is retained. Voice cloning and fine-tuning are excluded. General terms permit model improvement/training unless separately agreed or an opt-out is processed.

Read the service policy (opens in a new tab)

Caching

Caching can reuse earlier input processing. Availability, lifetime and billing depend on the endpoint. A cache hit still generates a new response.

01 / INITIAL INPUTProcess & retain

The system processes eligible input and retains reusable state.

02 / LATER REQUESTCheck available state

Reuse depends on matching input, session state and the cache policy.

03 / CACHE HITReuse & generate

Reusing state can reduce repeated work. Any charges follow the billing rules below.

Cartesia

ink-2
Not verified
How it works
Not collected
Minimum reusable input
Not collected
Lifetime & refresh
Not collected
How to enable it
Not collected
Usage fields
Not collected
Billing rules
Not collected
Limits & data boundaries
Not collected

No verified cache rate for this endpoint. An empty rate is not a zero charge.

Cartesia

ink-2
Not verified
How it works
Not collected
Minimum reusable input
Not collected
Lifetime & refresh
Not collected
How to enable it
Not collected
Usage fields
Not collected
Billing rules
Not collected
Limits & data boundaries
Not collected

No verified cache rate for this endpoint. An empty rate is not a zero charge.

What the measurements say

No benchmark observations have been added for this model. We do not turn a vendor description into a performance score.

Latency and throughput depend on the host, request size, reasoning settings and load. There is no measured speed or global usage figure in this record.

The record behind the model

Why it is included

Service terms prohibit downloading or distributing the hosted models and reserve provider model rights. These Sonic/Ink service variants are not in the reviewed official public-weight inventory. Classification is specific to the served variant: public Rene, Llamba, Mamba and H-Net weights are excluded; Ink Whisper is a provider-specific derivative, not a reclassification of baseline Whisper.

Classification evidence (opens in a new tab)

Lifecycle dates

Earlier launch announcement9 Jul 2026Source (opens in a new tab)

Sources reviewed