Output billingSee billing notesAccess and billing notes 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
No model-specific specification verified. Unknown does not mean unsupported.
Maximum input tokens
Not collected
No model-specific specification verified. Unknown does not mean unsupported.
Maximum output
Not collected
No model-specific specification verified. Unknown does not mean unsupported.
Parameters
Not collected
No model-specific specification verified. Unknown does not mean unsupported.
Release date
Not collected
No model-specific specification verified. Unknown does not mean unsupported.
Knowledge cutoff
Not collected
No model-specific specification verified. Unknown does not mean unsupported.
Tool calling
Not collected
No model-specific specification verified. Unknown does not mean unsupported.
Structured output
Not collected
No model-specific specification verified. Unknown does not mean unsupported.
Streaming
Not collected
No model-specific specification verified. Unknown does not mean unsupported.
Batch processing
Not collected
No model-specific specification verified. Unknown does not mean unsupported.
Fine-tuning
Not collected
No model-specific specification verified. Unknown does not mean unsupported.
Reasoning
Not collected
No model-specific specification verified. Unknown does not mean unsupported.
Context caching
Not collected
No model-specific specification verified. Unknown does not mean unsupported.
Base model
Not collected
No model-specific specification verified. Unknown does not mean unsupported.
Endpoint availability
LMNT has shut down. Historical contracts below do not establish a callable service. Exact shutdown date and any continuing third-party deployment are unverified.Source for Endpoint availability (opens in a new tab)
Creation privacy
Archived June 2023 policy allows user-content use for machine-learning training and human review. Personal information is generally retained until edited/deleted or while the person remains a user; no fixed inference-retention period or zero-retention promise established. US and other international transfers are described. Post-shutdown deletion and handling are unverified.Source for Creation privacy (opens in a new tab)
Model terms
Archived site terms describe personal, noncommercial site access; historical paid API plans advertise commercial licenses. These have different scopes, so neither a universal commercial prohibition nor unrestricted commercial rights is inferred. Separate service restrictions protect model weights. Current contractual status after shutdown is unknown.Source for Model terms (opens in a new tab)
5–10 seconds of representative reference speech recommended by archived guide. SDK accepts WAV, MP3, MP4, M4A and WebM; maximum upload 250 MB.Source for Reference audio (opens in a new tab)
File constraints independently documented by official voice-create SDK.
One reference file, maximum 250 MB, plus a name. Optional description, gender and tags. Gender is descriptive metadata and has no effect on voice creation.Source for Input limits (opens in a new tab)
Response format
Voice object with id, name, owner, state and optional metadata, preview_url and type=instant. ready/training are examples, not an exhaustive state enum. Use voice ID in a separate synthesis request.Source for Response format (opens in a new tab)
Stock voice catalog
Authenticated list supports system/me/all ownership and starred filtering. Voice profiles are user or catalogue prompts; complete private inventory was not accessed.Source for Stock voice catalog (opens in a new tab)
Model selection
A saved reference prompt is a managed profile, not a published independent base checkpoint. Professional profiles were converted to instant profiles in May 2025.Source for Model selection (opens in a new tab)
Published price
Historical plans include unlimited voice clones; downstream speech is billed by the applicable character subscription. No separate voice-enrollment price verified.Source for Published price (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.
No immutable model selector verified. Voice IDs identify profiles, not base checkpoints.
Access
LMNT has shut down. Historical contracts below do not establish a callable service. Exact shutdown date and any continuing third-party deployment are unverified. Historical POST https://api.lmnt.com/v1/ai/voice with name and multipart file. GET/PUT/DELETE /v1/ai/voice/{id} inspect, update metadata or delete/cancel pending operations. Returned voice profile is used by separate speech endpoints.
Regions
Historical inference residency unverified. Company location is not processing-region evidence.
Endpoint pricing has not been collected. Check the provider for applicable rates.
Data & enterprise terms
Archived June 2023 policy allows user-content use for machine-learning training and human review. Personal information is generally retained until edited/deleted or while the person remains a user; no fixed inference-retention period or zero-retention promise established. US and other international transfers are described. Post-shutdown deletion and handling are unverified. Archived site terms describe personal, noncommercial site access; historical paid API plans advertise commercial licenses. These have different scopes, so neither a universal commercial prohibition nor unrestricted commercial rights is inferred. Separate service restrictions protect model weights. Current contractual status after shutdown is unknown.
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.
LMNT API (retired)
Not verified
No inference-cache contract verified. Persistent speech sessions, stored voice prompts, saved clips, SDK cached properties and random seeds are not discounted input caching.
How it works
Not collected
No inference-cache contract verified. Persistent speech sessions, stored voice prompts, saved clips, SDK cached properties and random seeds are not discounted input caching.
Minimum reusable input
Not collected
No inference-cache contract verified. Persistent speech sessions, stored voice prompts, saved clips, SDK cached properties and random seeds are not discounted input caching.
Lifetime & refresh
Not collected
No inference-cache contract verified. Persistent speech sessions, stored voice prompts, saved clips, SDK cached properties and random seeds are not discounted input caching.
How to enable it
Not collected
No inference-cache contract verified. Persistent speech sessions, stored voice prompts, saved clips, SDK cached properties and random seeds are not discounted input caching.
Usage fields
Not collected
No inference-cache contract verified. Persistent speech sessions, stored voice prompts, saved clips, SDK cached properties and random seeds are not discounted input caching.
Billing rules
Not collected
No inference-cache contract verified. Persistent speech sessions, stored voice prompts, saved clips, SDK cached properties and random seeds are not discounted input caching.
Limits & data boundaries
Not collected
No inference-cache contract verified. Persistent speech sessions, stored voice prompts, saved clips, SDK cached properties and random seeds are not discounted input caching.
No verified cache rate for this endpoint. An empty rate is not a zero charge.
Caching reviewed 16 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
Saved voice-prompt enrollment and management is documented. No stable independent base-model checkpoint is identified; this managed component must not inflate the count of native private models.
