Skip to content
Model infoModelsSnapshot: 15 Oct 2026

Decart / Decart lip sync

LipSync Live

Retired managed lip-sync service combining streamed audio and portrait-video frames. Underlying checkpoint identity is not established.

retiredClosed-source managed componentLip syncAvatar animation
Official documentation (opens in a new tab)Reviewed 15 Oct 2026
8 of 22 specification fields collected
Output resolutionNot collectedAwaiting source review
Clip durationNot collectedAwaiting source review
Input billingNot collectedDecart API (retired)
Output billingNot collectedNo inferred price

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

Not established for this historical hosted model; no successor or unrelated provider capabilities inherited.

Maximum input tokens
Not collected

Not established for this historical hosted model; no successor or unrelated provider capabilities inherited.

Maximum output
Not collected

Not established for this historical hosted model; no successor or unrelated provider capabilities inherited.

Reasoning
Not collected

Not established for this historical hosted model; no successor or unrelated provider capabilities inherited.

Tool calling
Not collected

Not established for this historical hosted model; no successor or unrelated provider capabilities inherited.

Structured output
Not collected

Not established for this historical hosted model; no successor or unrelated provider capabilities inherited.

Batch processing
Not collected

Not established for this historical hosted model; no successor or unrelated provider capabilities inherited.

Fine-tuning
Not collected

Not established for this historical hosted model; no successor or unrelated provider capabilities inherited.

Knowledge cutoff
Not collected

Not established for this historical hosted model; no successor or unrelated provider capabilities inherited.

Release date
Not collected

Not established for this historical hosted model; no successor or unrelated provider capabilities inherited.

Parameters
Not collected

Not established for this historical hosted model; no successor or unrelated provider capabilities inherited.

Architecture
Not collected

Not established for this historical hosted model; no successor or unrelated provider capabilities inherited.

Languages
Not collected

Not established for this historical hosted model; no successor or unrelated provider capabilities inherited.

Context caching
Not collected

Not established for this historical hosted model; no successor or unrelated provider capabilities inherited.

Endpoint availability
Native service retired; notice appears in May 2026.Source for Endpoint availability (opens in a new tab)
Research publication
2025-10-22Source for Research publication (opens in a new tab)

Provider integration article; not asserted as the launch date.

Release window
2025-10Source for Release window (opens in a new tab)

Provider added realtime lip-sync capability in October; exact date not published.

Audio sample rates
Reference client default 16,000 HzSource for Audio sample rates (opens in a new tab)

Configurable integer; supported server range is not established.

Frame rate
Reference client default 25 FPSSource for Frame rate (opens in a new tab)

Configurable integer; not a maximum or output SLA.

Access & pricing

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

API model ID
Not collected

No named model selector in this dedicated service contract.

Access
Retired native service; retained for historical identity and integration context.
Regions
Not established
API endpoint URL
wss://api.decart.ai/router/lipsync/wsSource for API endpoint URL (opens in a new tab)
Authentication
Pinned reference client appends an API-key query parameter. This documents the old contract; never expose keys in frontend code or logs.Source for Authentication (opens in a new tab)
Input media transport
JSON WebSocket messages; binary audio and video fields are encoded as base64.Source for Input media transport (opens in a new tab)
Required request fields
config.video_fps, config.audio_sample_rate, audio_input.audio_data, video_input.video_frameSource for Required request fields (opens in a new tab)

Fields belong to distinct message types, not a single request.

Live session controls
config, config_ack, audio_input, video_input, interrupt_audio, synced_result, errorSource for Live session controls (opens in a new tab)
Source audio handling
interrupt_audio flushes pending speech; synchronized output pairs video_frame and audio_frame. External TTS supplies speech, so this is not a Decart text-to-speech model.Source for Source audio handling (opens in a new tab)
Output formats
Provider article describes JPEG or RGB24 video frames; exact negotiated server format constraints are not established.Source for Output formats (opens in a new tab)
Input limits
Not collected

No numeric FPS/sample-rate range, frame-size ceiling or session maximum in the pinned schema.

Model selection
Not collected

No model selector identifies the backing checkpoint; do not equate this service with LiveAvatar or a third-party lip-sync model.

Endpoint availability
Provider May 2026 changelog says no longer served.Source for Endpoint availability (opens in a new tab)
Pricing coverage
Not collected

No verified historical tariff. Current Lucy, avatar or lip-sync prices must not be substituted.

Endpoint pricing has not been collected. Check the provider for applicable rates.

Data & enterprise terms

Retention, training use, regional processing and enterprise terms have not been summarized for this endpoint.

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.

Decart API (retired)

Not verified

Historical inference caching and cache billing are not established. Streaming buffers and uploaded reference files are separate mechanisms.

How it works
Not collected

Historical inference caching and cache billing are not established. Streaming buffers and uploaded reference files are separate mechanisms.

Minimum reusable input
Not collected

Historical inference caching and cache billing are not established. Streaming buffers and uploaded reference files are separate mechanisms.

Lifetime & refresh
Not collected

Historical inference caching and cache billing are not established. Streaming buffers and uploaded reference files are separate mechanisms.

How to enable it
Not collected

Historical inference caching and cache billing are not established. Streaming buffers and uploaded reference files are separate mechanisms.

Usage fields
Not collected

Historical inference caching and cache billing are not established. Streaming buffers and uploaded reference files are separate mechanisms.

Billing rules
Not collected

Historical inference caching and cache billing are not established. Streaming buffers and uploaded reference files are separate mechanisms.

Limits & data boundaries
Not collected

Historical inference caching and cache billing are not established. Streaming buffers and uploaded reference files are separate mechanisms.

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

Caching reviewed 15 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

The provider describes a hosted lip-sync service, but the underlying model identity and weight provenance are not established. This is a managed service record, not an additional claimed private foundation checkpoint.

Classification evidence (opens in a new tab)

Sources reviewed