Skip to content
Model infoModelsSnapshot: 15 Oct 2026

Sanas / Accent Translation

Accent Translation AT5.2

Streaming accent conversion with US and UK output accents, exposed through the developer SDK.

availableClosed weightsAudio processingSpeech-to-speech
Official documentation (opens in a new tab)Reviewed 13 Oct 2026
8 of 20 specification fields collected
Audio sample ratesUp to 24 kHzSee specification notes below
Audio formatsInterleaved float32 PCM through the Python SDKSee specification notes below
Input billingSee ratesSanas Cloud · Python SDK
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
Reasoning
Not collected
Tool calling
Not collected
Structured output
Not collected
Streaming
SupportedSource for Streaming (opens in a new tab)

Streaming speech processing, not LLM text-token streaming.

Batch processing
Not collected
Fine-tuning
Not collected
Knowledge cutoff
Not collected
Release date
Not collected
Parameters
Not collected
Architecture
Not collected
Context caching
Not collected
latency
Approximately 200 ms in the model overviewSource for latency (opens in a new tab)

Provider-reported processing latency, not end-to-end network latency or a guaranteed SLA.

Audio formats
Interleaved float32 PCM through the Python SDKSource for Audio formats (opens in a new tab)
Output accents
US English, UK EnglishSource for Output accents (opens in a new tab)

Access & pricing

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

Sanas Cloud · Python SDK

Provider source (opens in a new tab)
API model ID
AT5.2

SDK model key, not a REST language-model identifier.

Access
Create and activate an SDK instance, choose AudioAttributes.model_name, wait for RUNNING, then process interleaved float32 PCM frames at realtime speed.
Regions
Sanas Cloud: AWS Ohio us-east-2 and Mumbai ap-south-1; select region through account/project configuration
Audio formats
Interleaved float32 PCM framesSource for Audio formats (opens in a new tab)
Streaming audio chunks
Example feeds 20 ms frames every 20 ms; source must be paced in real timeSource for Streaming audio chunks (opens in a new tab)

An integration contract/example, not a maximum audio or context duration.

Request quotas
Not collected

No hard per-account concurrency or request cap verified. Marketing claims of hundreds of streams are not an entitlement.

Billing basis
Prepaid wallet; metered minutes deducted in real timeSource for Billing basis (opens in a new tab)
Included free allowance
$10 introductory credit, with expiry shown in the account walletSource for Included free allowance (opens in a new tab)

Promotional allowance, not a zero-dollar ongoing model rate.

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 13 Oct 2026
CategoryUsageRateUnit
Audio processingCloud inference1$0.009processed audio minute

Rate conditions

  1. Published Sanas Cloud developer rate, metered from a prepaid wallet. Volume discounts by agreement. Desktop subscriptions, self-hosting and partner contracts are separate. Rounding and simultaneous-direction billing are not specified here.
Pricing source (opens in a new tab)

Data & enterprise terms

Developer documentation says live call audio is processed without cloud audio storage. Account, usage and other personal data have contract-based retention: six-month post-termination deletion and up to one year for backups, with 30 days to request return/deletion. This is not a promise that all metadata is discarded. Consented samples and diagnostics require separate review.

Read the service policy (opens in a new tab)

Sanas Cloud · Browser SDK

Provider source (opens in a new tab)
API model ID
AT5.2

SDK model key, not a REST language-model identifier.

Access
Remote audio processor with modelName matching the Developer Console. Browser uses a short-lived token minted by the application backend; register allowed origins.
Regions
Sanas Cloud: AWS Ohio us-east-2 and Mumbai ap-south-1; select region through account/project configuration
Audio formats
Browser audio tracks; SDK performs remote audio processingSource for Audio formats (opens in a new tab)
Streaming audio chunks
SDK-managed stream and configurable jitter bufferSource for Streaming audio chunks (opens in a new tab)

An integration contract/example, not a maximum audio or context duration.

Request quotas
Not collected

No hard per-account concurrency or request cap verified. Marketing claims of hundreds of streams are not an entitlement.

Billing basis
Prepaid wallet; metered minutes deducted in real timeSource for Billing basis (opens in a new tab)
Included free allowance
$10 introductory credit, with expiry shown in the account walletSource for Included free allowance (opens in a new tab)

Promotional allowance, not a zero-dollar ongoing model rate.

Audio sample rates
Browser SDK accepts 8, 16 or 48 kHz; default 16 kHzSource for Audio sample rates (opens in a new tab)

Transport sample rates do not change a model’s native bandwidth or guarantee native 48 kHz inference.

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 13 Oct 2026
CategoryUsageRateUnit
Audio processingCloud inference1$0.009processed audio minute

Rate conditions

  1. Published Sanas Cloud developer rate, metered from a prepaid wallet. Volume discounts by agreement. Desktop subscriptions, self-hosting and partner contracts are separate. Rounding and simultaneous-direction billing are not specified here.
Pricing source (opens in a new tab)

Data & enterprise terms

Developer documentation says live call audio is processed without cloud audio storage. Account, usage and other personal data have contract-based retention: six-month post-termination deletion and up to one year for backups, with 30 days to request return/deletion. This is not a promise that all metadata is discarded. Consented samples and diagnostics require separate review.

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.

Sanas Cloud · Python SDK

AT5.2
Not verified

No model-specific discounted prompt-cache contract published in the reviewed sources. Audio buffering, model loading and session state are not treated as token caching.

How it works
Not collected

No model-specific discounted prompt-cache contract published in the reviewed sources. Audio buffering, model loading and session state are not treated as token caching.

Minimum reusable input
Not collected

No model-specific discounted prompt-cache contract published in the reviewed sources. Audio buffering, model loading and session state are not treated as token caching.

Lifetime & refresh
Not collected

No model-specific discounted prompt-cache contract published in the reviewed sources. Audio buffering, model loading and session state are not treated as token caching.

How to enable it
Not collected

No model-specific discounted prompt-cache contract published in the reviewed sources. Audio buffering, model loading and session state are not treated as token caching.

Usage fields
Not collected

No model-specific discounted prompt-cache contract published in the reviewed sources. Audio buffering, model loading and session state are not treated as token caching.

Billing rules
Not collected

No model-specific discounted prompt-cache contract published in the reviewed sources. Audio buffering, model loading and session state are not treated as token caching.

Limits & data boundaries
Not collected

No model-specific discounted prompt-cache contract published in the reviewed sources. Audio buffering, model loading and session state are not treated as token caching.

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

Sanas Cloud · Browser SDK

AT5.2
Not verified

No model-specific discounted prompt-cache contract published in the reviewed sources. Audio buffering, model loading and session state are not treated as token caching.

How it works
Not collected

No model-specific discounted prompt-cache contract published in the reviewed sources. Audio buffering, model loading and session state are not treated as token caching.

Minimum reusable input
Not collected

No model-specific discounted prompt-cache contract published in the reviewed sources. Audio buffering, model loading and session state are not treated as token caching.

Lifetime & refresh
Not collected

No model-specific discounted prompt-cache contract published in the reviewed sources. Audio buffering, model loading and session state are not treated as token caching.

How to enable it
Not collected

No model-specific discounted prompt-cache contract published in the reviewed sources. Audio buffering, model loading and session state are not treated as token caching.

Usage fields
Not collected

No model-specific discounted prompt-cache contract published in the reviewed sources. Audio buffering, model loading and session state are not treated as token caching.

Billing rules
Not collected

No model-specific discounted prompt-cache contract published in the reviewed sources. Audio buffering, model loading and session state are not treated as token caching.

Limits & data boundaries
Not collected

No model-specific discounted prompt-cache contract published in the reviewed sources. Audio buffering, model loading and session state are not treated as token caching.

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

Sanas documents these as its trained speech models and supplies them through licensed applications and hosted SDK inference. Its terms reserve the software/content rights and prohibit reverse engineering and redistribution. No open-weight release is established by these primary model and access documents.

Classification evidence (opens in a new tab)

Sources reviewed