Skip to content
Model infoModelsSnapshot: 15 Oct 2026

Resemble AI / Resemble detection and provenance

DETECT-World

Combines learned generation-artifact signals with physical-consistency checks for synthetic audio, images and video.

availableClosed weightsDeepfake detection
Official documentation (opens in a new tab)Reviewed 13 Oct 2026
10 of 20 specification fields collected
Direct file limit150 MB direct media upload; larger files use secure uploadSee specification notes below
Batch limitUp to 50 files, 500 MB total; asynchronous onlySee specification notes below
Input billingSee ratesResemble AI
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
Input modalities
Audio, Image, VideoSource for Input modalities (opens in a new tab)
Output modalities
Classification scoresSource for Output modalities (opens in a new tab)
Reasoning
Not collected
Tool calling
Not collected
Structured output
Not collected
Streaming
SupportedSource for Streaming (opens in a new tab)

Audio-only streaming. No live image/video stream contract established.

Fine-tuning
Not collected
Knowledge cutoff
Not collected
Parameters
Not collected
Architecture
DETECT-3B Omni foundation plus a physical-consistency detection layerSource for Architecture (opens in a new tab)

The full system parameter count is not supplied by the reviewed source; do not reuse the foundation’s 3B count.

Languages
Not collected
Context caching
Not collected
Batch limits
Up to 50 files, 500 MB total; asynchronous onlySource for Batch limits (opens in a new tab)
Maximum audio file
150 MB direct media upload; larger files use secure uploadSource for Maximum audio file (opens in a new tab)

The limit applies to all supported direct-upload media. Optional watermark decoding has a separate 200 MB limit.

Published price
Flex: audio $0.035/second; image $0.035/image; video $0.07/secondSource for Published price (opens in a new tab)
Usage metering
Team/Business: audio $0.015/second, image $0.015/image, video $0.03/second; subscription fee additionalSource for Usage metering (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.

API model ID
Not collected

No public request selector pins this model version.

Access
API access and billing eligibility required. Single-media submission with optional waiting, polling or callback.
Regions
Region and data residency not established for this access path.
API endpoint URL
https://app.resemble.ai/api/v2/detectSource for API endpoint URL (opens in a new tab)
Request mode
Asynchronous by default; Prefer: wait supported for single-file submissionsSource for Request mode (opens in a new tab)
Input modalities
Audio, Image, VideoSource for Input modalities (opens in a new tab)
Output modalities
Classification scoresSource for Output modalities (opens in a new tab)
Maximum audio file
150 MB direct media upload; larger files use secure uploadSource for Maximum audio file (opens in a new tab)

The limit applies to all supported direct-upload media. Optional watermark decoding has a separate 200 MB limit.

Generation controls
Optional face-only visual analysis, modality selection, regional analysis and visualizationsSource for Generation controls (opens in a new tab)

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 13 Oct 2026
CategoryUsageRateUnit
Detection / provenanceAudio detection · Flex1$0.035second of audio
Detection / provenanceImage detection · Flex1$0.035image
Detection / provenanceVideo detection · Flex1$0.07second of video
Detection / provenanceAudio detection · Team/Business2$0.015second of audio
Detection / provenanceImage detection · Team/Business2$0.015image
Detection / provenanceVideo detection · Team/Business2$0.03second of video

Rate conditions

  1. No base subscription fee; usage credits required. Published current Detect service rate, not a historical DETECT-2B/3B quote. Optional Intelligence, watermark and other products bill separately.
  2. Paid subscription fee additional. Published current Detect service rate, not a historical DETECT-2B/3B quote. Optional Intelligence, watermark and other products bill separately.
Pricing source (opens in a new tab)

Data & enterprise terms

Terms distinguish voice Content from Detection Media. Detection Media is not used for model training without express written opt-in. Deletion/retention follow service policies and any applicable DPA; no unconditional zero-retention guarantee.

Read the service policy (opens in a new tab)
API model ID
Not collected

No public request selector pins this model version.

Access
Batch upload requires an eligible plan; the pricing table starts batch uploads at Team.
Regions
Region and data residency not established for this access path.
API endpoint URL
https://app.resemble.ai/api/v2/detect/batchSource for API endpoint URL (opens in a new tab)
Batch limits
Up to 50 files, 500 MB total; asynchronous onlySource for Batch limits (opens in a new tab)
Endpoint caveat
Exactly one of files[] or a ZIP archive. Prefer: wait and watermark decoding rejected.Source for Endpoint caveat (opens in a new tab)

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

Data & enterprise terms

Terms distinguish voice Content from Detection Media. Detection Media is not used for model training without express written opt-in. Deletion/retention follow service policies and any applicable DPA; no unconditional zero-retention guarantee.

Read the service policy (opens in a new tab)
API model ID
Not collected

No public request selector pins this model version.

Access
Server-side authenticated WebSocket with detection entitlement and available balance.
Regions
Region and data residency not established for this access path.
API endpoint URL
wss://stream.resemble.ai/api/v1/detect/audioSource for API endpoint URL (opens in a new tab)
Audio formats
WAV stream with signed 16-bit little-endian PCMSource for Audio formats (opens in a new tab)
Required request fields
Begin with a standard 44-byte WAV header, followed by binary PCM framesSource for Required request fields (opens in a new tab)
Audio duration limits
Approximately 4-second analysis windowsSource for Audio duration limits (opens in a new tab)

Window length, not a session-duration limit.

Endpoint caveat
Insufficient voice activity yields skipped windows. Stream results are not persisted as Detect jobs. File-route callback and visualization options are not accepted.Source for Endpoint caveat (opens in a new tab)

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

Data & enterprise terms

Terms distinguish voice Content from Detection Media. Detection Media is not used for model training without express written opt-in. Deletion/retention follow service policies and any applicable DPA; no unconditional zero-retention guarantee.

Read the service policy (opens in a new tab)
API model ID
Not collected

No public request selector pins this model version.

Access
Enterprise on-premises and air-gapped deployment advertised. Quote and contract required; no public downloadable weights inferred.
Regions
Region and data residency not established for this access path.

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

Data & enterprise terms

Terms distinguish voice Content from Detection Media. Detection Media is not used for model training without express written opt-in. Deletion/retention follow service policies and any applicable DPA; no unconditional zero-retention guarantee.

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.

Resemble AI

Not 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

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

Resemble AI

Not 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

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

Resemble AI

Not 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

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

Resemble AI

Not 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

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

Provider documentation establishes the named native model or managed composition. Terms expressly reserve proprietary voice/detection technology and prohibit reverse engineering of models. Public-weight Chatterbox, original PerTh and Resemblyzer are excluded; enterprise on-premises delivery is not treated as a public weight release.

Classification evidence (opens in a new tab)

Lifecycle dates

Official DETECT-World introduction12 Aug 2026Source (opens in a new tab)

Sources reviewed