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 input tokens
Not collected
Maximum output
Not collected
Reasoning
Not collected
Tool calling
Not collected
Structured output
Not collected
Streaming
Not collected
Batch processing
Not collected
Fine-tuning
Not collected
A custom branded voice is documented, but its training mechanism and customer-accessible tuning workflow were not established.
Knowledge cutoff
Not collected
Release date
Not collected
Parameters
Not collected
Languages
Not collected
Context caching
Not collected
Endpoint availability
Custom voice provisioning required. The API docs reference YourVoice, but its product URL currently resolves to the portal homepage; current sales availability and custom contract terms remain unverified.Source for Endpoint availability (opens in a new tab)
The API selects a voice rather than publishing an immutable shared foundation-model ID.
Access
Custom branded voice referenced by the SaluteSpeech synthesis API. Requires provider provisioning and a customer voice ID; current sales availability not verified.
Regions
Processing region not verified for the exact account and deployment.
Endpoint pricing has not been collected. Check the provider for applicable rates.
Data & enterprise terms
Audio retention, model-training use and account-specific residency terms not established in this review.
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.
Sber SaluteSpeech
Not verified
No verified model-specific cache TTL, discount or hit-accounting policy. GigaChat text caching does not apply by inference.
How it works
Not collected
No verified model-specific cache TTL, discount or hit-accounting policy. GigaChat text caching does not apply by inference.
Minimum reusable input
Not collected
No verified model-specific cache TTL, discount or hit-accounting policy. GigaChat text caching does not apply by inference.
Lifetime & refresh
Not collected
No verified model-specific cache TTL, discount or hit-accounting policy. GigaChat text caching does not apply by inference.
How to enable it
Not collected
No verified model-specific cache TTL, discount or hit-accounting policy. GigaChat text caching does not apply by inference.
Usage fields
Not collected
No verified model-specific cache TTL, discount or hit-accounting policy. GigaChat text caching does not apply by inference.
Billing rules
Not collected
No verified model-specific cache TTL, discount or hit-accounting policy. GigaChat text caching does not apply by inference.
Limits & data boundaries
Not collected
No verified model-specific cache TTL, discount or hit-accounting policy. GigaChat text caching does not apply by inference.
No verified cache rate for this endpoint. An empty rate is not a zero charge.
Caching reviewed 14 Oct 2026
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-documented hosted speech component. Public API schemas do not publish the served weights. The exact underlying checkpoint and its relationship to public GigaAM releases are not established, so this record does not assert a separate proprietary foundation model.
