Skip to content
Model infoModelsSnapshot: 15 Oct 2026

Yandex / Speech Realtime

Speech Realtime 250923

Speech Realtime 250923: voice conversation service, retained for historical reference.

Speech & voice
retiredClosed-source managed componentVoice conversation
Official documentation (opens in a new tab)Reviewed 14 Oct 2026
6 of 18 specification fields collected
Context window32.8K32,768 tokens
Maximum outputNot collectedAwaiting source review
Input billingSee ratesYandex AI Studio (retired)
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
32,768 tokensSource for Context / input limit (opens in a new tab)

Shared context window, not independent input plus output allowances.

Maximum input tokens
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)

Realtime API voice sessions.

Batch processing
Not collected
Fine-tuning
Not collected
Knowledge cutoff
Not collected
Release date
Not collected
Parameters
Not collected
Languages
Not collected
Context caching
Not collected

Retired selector; historical text cache rates are retained separately.

Input modalities
Text, AudioSource for Input modalities (opens in a new tab)
Output modalities
AudioSource for Output modalities (opens in a new tab)
Native model options
gpt://<folder_ID>/speech-realtime-250923Source for Native model options (opens in a new tab)
Billing basis
Text tokens, audio input, audio output and tools are separately billable.Source for Billing basis (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.

Yandex AI Studio (retired)

Provider source (opens in a new tab)
API model ID
gpt://<folder_ID>/speech-realtime-250923
Access
Retired; not a current callable endpoint.
Regions
Contract and deployment dependent; exact processing location not verified.

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 14 Oct 2026
CategoryUsageRateUnit
Historical voice pipeline ratesText input1$6.5573761M tokens
Historical voice pipeline ratesText output1$6.5573761M tokens
Historical voice pipeline ratesCached text input1$1.6393441M tokens
Historical voice pipeline ratesBuilt-in tool text1$1.6393441M tokens
Historical voice pipeline ratesIncoming audio2$0.00021639second
Historical voice pipeline ratesOutgoing audio3$0.00016639second

Rate conditions

  1. Published USD rates for Iron Hive doo Beograd (Serbia) or Direct Cursus Technology L.L.C. (Dubai) contracts; VAT excluded. Not a conversion of RUB or KZT rates. Only the text component; audio and tool invocation charges are additional. Selector switched off on 12 October 2026; not a current quote.
  2. Published USD rates for Iron Hive doo Beograd (Serbia) or Direct Cursus Technology L.L.C. (Dubai) contracts; VAT excluded. Not a conversion of RUB or KZT rates. Add text, output audio and any tool charges.
  3. Published USD rates for Iron Hive doo Beograd (Serbia) or Direct Cursus Technology L.L.C. (Dubai) contracts; VAT excluded. Not a conversion of RUB or KZT rates. LiveTTS has a different outgoing-audio rate from general synthesis.
Pricing source (opens in a new tab)

Data & enterprise terms

Model-specific retention and training-use terms not collected.

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.

Yandex AI Studio (retired)

gpt://<folder_ID>/speech-realtime-250923
Not verified

No current exact-endpoint cache policy 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
Cache reads, writes & storage
ChargeApplies toRate / unit
Cached text input1Historical voice pipeline rates$1.639344 / 1M tokensPricing evidence (opens in a new tab)

Rate conditions

  1. Published USD rates for Iron Hive doo Beograd (Serbia) or Direct Cursus Technology L.L.C. (Dubai) contracts; VAT excluded. Not a conversion of RUB or KZT rates. Only the text component; audio and tool invocation charges are additional. Selector switched off on 12 October 2026; not a current quote.

Observed 14 Oct 2026. Write rates and storage fees are different charges; use the billing rules above.

Caching reviewed 14 Oct 2026

Voice choices

28 voices

Choose a voice within Speech Realtime 250923. These are API voice identifiers, not additional model releases. Source links document each voice.

Published voice catalog
Voice / API IDLanguage & accentDelivery
Lealeade-DE
Not separately documented		SpeechKit general voice. No selectable role listed. API v1, v3 and compatible general-voice Realtime models.Voice source (opens in a new tab)
Johnjohnen-US
Not separately documented		SpeechKit general voice. No selectable role listed. API v1, v3 and compatible general-voice Realtime models.Voice source (opens in a new tab)
Naominaomihe-IL
Not separately documented		SpeechKit general voice. Roles: modern, classic. API v3 and compatible general-voice Realtime models.Voice source (opens in a new tab)
Amiraamirakk-KK или kk-KZ
Not separately documented		SpeechKit general voice. No selectable role listed. API v1, v3 and compatible general-voice Realtime models.Voice source (opens in a new tab)
Madimadikk-KK или kk-KZ
Not separately documented		SpeechKit general voice. No selectable role listed. API v1, v3 and compatible general-voice Realtime models.Voice source (opens in a new tab)
Saulesaulekk-KK или kk-KZ
Not separately documented		SpeechKit general voice. Roles: neutral, strict. API v3 and compatible general-voice Realtime models.Voice source (opens in a new tab)
Zhanarzhanarkk-KK или kk-KZ
Not separately documented		SpeechKit general voice. Roles: neutral, friendly. API v3 and compatible general-voice Realtime models.Voice source (opens in a new tab)
Filippfilippru-RU
Not separately documented		SpeechKit general voice. No selectable role listed. API v1, v3 and compatible general-voice Realtime models.Voice source (opens in a new tab)

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

Provider-hosted classification, custom deployment template or speech pipeline. The complete served component is not downloadable in the reviewed documentation; this does not assert private weights for every backbone or a new foundation checkpoint.

Classification evidence (opens in a new tab)

Lifecycle dates

Service version switched off / support ended12 Oct 2026Source (opens in a new tab)

Sources reviewed