Skip to content
Model infoModelsSnapshot: 15 Oct 2026

Reson8 / Resonant

Reson8 Speech-to-Text · managed family

Current production speech pipeline for prerecorded files, live transcripts and turn-aware voice agents. Updated server-side without a public base-model version selector.

Transcription
availableClosed weightsSpeech recognition
Official documentation (opens in a new tab)Reviewed 13 Oct 2026
13 of 24 specification fields collected
Output formatsJSON transcripts, optional word details and turn events, depending on endpointSee specification notes below
Languages10 listedFull language list in specifications
Input billingSee ratesReson8 API · prerecorded
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
AudioSource for Input modalities (opens in a new tab)
Output modalities
TextSource for Output modalities (opens in a new tab)
Reasoning
Not collected
Tool calling
Not collected
Structured output
Not collected

Fixed JSON responses and regex-like transcript recovery are documented, not caller-defined constrained generative JSON schemas.

Streaming
SupportedSource for Streaming (opens in a new tab)
Batch processing
SupportedSource for Batch processing (opens in a new tab)

Synchronous whole-file transcription; not an asynchronous multi-job batch queue.

Fine-tuning
Not collected

Provider describes no-fine-tuning terminology adaptation via proprietary context-biasing adapters. This does not establish a general user weight fine-tuning API.

Knowledge cutoff
Not collected
Release date
Not collected
Parameters
Not collected
Architecture
Not collected
Languages
German (de), English (en), Spanish (es), French (fr), Frisian (fy), Italian (it), Dutch (nl), Polish (pl), Portuguese (pt), Swedish (sv)Source for Languages (opens in a new tab)
Context caching
Not collected

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

Language switching within audio
SupportedSource for Language switching within audio (opens in a new tab)

Omit language for per-utterance detection, pin one language for best quality, or restrict detection with a comma-separated list. Short utterances can be less reliable.

Output formats
JSON transcripts, optional word details and turn events, depending on endpointSource for Output formats (opens in a new tab)
Audio formats
WAV, M4A, MP3, OGG, FLAC, WebM, Raw PCM s16le, G.711 μ-law, G.711 A-lawSource for Audio formats (opens in a new tab)

Container auto-detection is default. Prerecorded also accepts explicit seekable media containers; fresh streaming connections need headers or explicit raw encoding.

Custom vocabulary
Up to 250 per-request phrases or 50,000 phrases per named custom model; no stated limit on number of custom modelsSource for Custom vocabulary (opens in a new tab)

Query custom_model_id overrides the API client default. bias_strength defaults 0.45 and is non-negative. Larger bias can introduce irrelevant terms.

Usage metering
Credits per audio minute: prerecorded 1; realtime or Turns 2; optional diarization +0.2.Source for Usage metering (opens in a new tab)
Subscription plans
Free: 600 credits/month and 1 connection. Pro: €20/month, 6,000 credits and 20 connections. Growth: €100/month, 30,000 credits and 100 connections. Enterprise: negotiated.Source for Subscription plans (opens in a new tab)

Included credits are shared account usage, not a separate free allowance on every endpoint. Unused-credit rollover and exact rounding are unverified.

Research controls
Patterns recover expected structured identifiers: up to 32 expanded characters excluding separators and 32 alternative shapesSource for Research controls (opens in a new tab)

Patterns cannot combine with phrases, custom_model_id, or an API-client default custom model. They may worsen transcription when the expected token is absent.

Deployment targets
Hosted Netherlands API; negotiated on-premises or VPC deploymentSource for Deployment targets (opens in a new tab)

Customer story demonstrates AWS EKS installed through Helm. Exact private checkpoint IDs and system requirements remain unverified.

Access & pricing

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

Reson8 API · prerecorded

Provider source (opens in a new tab)
API model ID
Not collected

API path selects a task/transport; custom_model_id selects a terminology adapter, not an immutable base-model version.

Access
POST https://api.reson8.dev/v1/speech-to-text/prerecorded with Authorization: ApiKey or Bearer, Content-Type: application/octet-stream, and the audio file as the raw request body. Receive a synchronous JSON transcript.
Regions
Netherlands · provider-owned GPU infrastructure
Languages
German (de), English (en), Spanish (es), French (fr), Frisian (fy), Italian (it), Dutch (nl), Polish (pl), Portuguese (pt), Swedish (sv)Source for Languages (opens in a new tab)
Streaming
Not supportedSource for Streaming (opens in a new tab)
Batch processing
SupportedSource for Batch processing (opens in a new tab)
Audio formats
WAV, M4A, MP3, OGG, FLAC, WebM, Raw PCM s16le, G.711 μ-law, G.711 A-lawSource for Audio formats (opens in a new tab)

Container auto-detection is default. Prerecorded also accepts explicit seekable media containers; fresh streaming connections need headers or explicit raw encoding.

Audio channels
1–10 input channels; default 1Source for Audio channels (opens in a new tab)

Transport channel count is distinct from max_speakers, whose maximum is eight on diarization endpoints.

Audio sample rates
Default 16,000 Hz when used with explicit encoding; exact full range unverifiedSource for Audio sample rates (opens in a new tab)

Recommended raw streaming input is 16 kHz mono PCM s16le.

Custom vocabulary
Up to 250 per-request phrases or 50,000 phrases per named custom model; no stated limit on number of custom modelsSource for Custom vocabulary (opens in a new tab)

Query custom_model_id overrides the API client default. bias_strength defaults 0.45 and is non-negative. Larger bias can introduce irrelevant terms.

Research controls
Patterns recover expected structured identifiers: up to 32 expanded characters excluding separators and 32 alternative shapesSource for Research controls (opens in a new tab)

Patterns cannot combine with phrases, custom_model_id, or an API-client default custom model. They may worsen transcription when the expected token is absent.

API request fields
language, phrases or custom_model_id, bias_strength, include_timestamps, include_words, include_language, include_confidence, patterns and filler_modeSource for API request fields (opens in a new tab)

Not all combinations are valid. Filler mode is clean, natural (default) or verbatim; realtime interim text is unaffected.

Speaker diarization
SupportedSource for Speaker diarization (opens in a new tab)

diarize=true adds anonymous session-local speaker labels. max_speakers accepts 1–8; when absent, count is automatic. Realtime reports the dominant speaker of a transcript.

Word timestamps
Optional include_words and include_timestamps; include_confidence adds word confidenceSource for Word timestamps (opens in a new tab)

With diarization, words and language move from the top-level response into speaker segments.

Concurrent requests
Shared active-session limits: Free 1, Pro 20, Growth 100, Enterprise negotiatedSource for Concurrent requests (opens in a new tab)

A prerecorded in-flight request or open realtime/Turns connection each consumes one slot. Limits are checked at request start; sessions in progress complete.

Billing basis
1 credits per audio minute; diarization adds 0.2 credits per minuteSource for Billing basis (opens in a new tab)

Free accounts stop accepting new requests once the credit limit is reached. Paid plans have no credit cap, but concurrency limits still apply.

Session duration
Not collected

No maximum session duration established by reviewed docs. A 30-second ping interval is a keepalive cadence, not a session limit.

Maximum audio file
Not collected

No numeric upload ceiling established by the reviewed prerecorded contract.

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 13 Oct 2026
CategoryUsageRateUnit
Speech recognitionPro prerecorded overage1€0.006audio minute
Speech recognition add-onPro diarization add-on overage2€0.0012audio minute
Speech recognitionGrowth prerecorded overage3€0.004audio minute
Speech recognition add-onGrowth diarization add-on overage4€0.0008audio minute
Speech recognitionPro account subscription5€20.00account month
Speech recognitionGrowth account subscription6€100.00account month

Rate conditions

  1. Pro costs €20/month including 6,000 shared credits. This overage rate applies only after included credits: 1 credits/minute × €0.006/credit. It is not a standalone pay-as-you-go offer or an all-in average. No USD conversion or model-release pinning is implied. Exact rounding and tax treatment require the account terms.
  2. Pro overage: 0.2 extra credits/minute × €0.006/credit. Additive to base transcription after shared included credits are exhausted. Requires the corresponding paid subscription; no diarization on Turns.
  3. Growth costs €100/month including 30,000 shared credits. This overage rate applies only after included credits: 1 credits/minute × €0.004/credit. It is not a standalone pay-as-you-go offer or an all-in average. No USD conversion or model-release pinning is implied. Exact rounding and tax treatment require the account terms.
  4. Growth overage: 0.2 extra credits/minute × €0.004/credit. Additive to base transcription after shared included credits are exhausted. Requires the corresponding paid subscription; no diarization on Turns.
  5. Account-wide subscription includes 6,000 credits shared across transports and applicable add-ons. Recorded once on the prerequisite prerecorded route for display; not an additional subscription charge per endpoint. Overage rates are separate.
  6. Account-wide subscription includes 30,000 credits shared across transports and applicable add-ons. Recorded once on the prerequisite prerecorded route for display; not an additional subscription charge per endpoint. Overage rates are separate.
Pricing source (opens in a new tab)

Data & enterprise terms

Hosted audio is processed on Reson8-owned infrastructure in the Netherlands, with no audio subprocessors according to the security guide. Audio is not retained after processing unless explicitly agreed otherwise. Terms retain terminology lists for the agreement and delete them within 30 days after termination, subject to legal exceptions. Audio and terminology are not used for general-model training without separate informed consent. Zero audio retention does not mean no stored account or terminology data; the website privacy policy has a different scope.

Read the service policy (opens in a new tab)

Reson8 API · realtime

Provider source (opens in a new tab)
API model ID
Not collected

API path selects a task/transport; custom_model_id selects a terminology adapter, not an immutable base-model version.

Access
Connect to wss://api.reson8.dev/v1/speech-to-text/realtime. Server clients use Authorization: ApiKey or Bearer. Browser clients use a short-lived bearer token in Sec-WebSocket-Protocol. Send binary audio and receive JSON text events.
Regions
Netherlands · provider-owned GPU infrastructure
Languages
German (de), English (en), Spanish (es), French (fr), Frisian (fy), Italian (it), Dutch (nl), Polish (pl), Portuguese (pt), Swedish (sv)Source for Languages (opens in a new tab)
Streaming
SupportedSource for Streaming (opens in a new tab)
Batch processing
Not supportedSource for Batch processing (opens in a new tab)
Audio formats
WAV, M4A, MP3, OGG, FLAC, WebM, Raw PCM s16le, G.711 μ-law, G.711 A-lawSource for Audio formats (opens in a new tab)

Container auto-detection is default. Prerecorded also accepts explicit seekable media containers; fresh streaming connections need headers or explicit raw encoding.

Audio channels
1–10 input channels; default 1Source for Audio channels (opens in a new tab)

Transport channel count is distinct from max_speakers, whose maximum is eight on diarization endpoints.

Audio sample rates
Default 16,000 Hz when used with explicit encoding; exact full range unverifiedSource for Audio sample rates (opens in a new tab)

Recommended raw streaming input is 16 kHz mono PCM s16le.

Custom vocabulary
Up to 250 per-request phrases or 50,000 phrases per named custom model; no stated limit on number of custom modelsSource for Custom vocabulary (opens in a new tab)

Query custom_model_id overrides the API client default. bias_strength defaults 0.45 and is non-negative. Larger bias can introduce irrelevant terms.

Research controls
Patterns recover expected structured identifiers: up to 32 expanded characters excluding separators and 32 alternative shapesSource for Research controls (opens in a new tab)

Patterns cannot combine with phrases, custom_model_id, or an API-client default custom model. They may worsen transcription when the expected token is absent.

API request fields
language, phrases or custom_model_id, bias_strength, include_timestamps, include_words, include_language, include_confidence, patterns and filler_modeSource for API request fields (opens in a new tab)

Not all combinations are valid. Filler mode is clean, natural (default) or verbatim; realtime interim text is unaffected.

Speaker diarization
SupportedSource for Speaker diarization (opens in a new tab)

diarize=true adds anonymous session-local speaker labels. max_speakers accepts 1–8; when absent, count is automatic. Realtime reports the dominant speaker of a transcript.

Word timestamps
Optional include_words and include_timestamps; include_confidence adds word confidenceSource for Word timestamps (opens in a new tab)

Detected language is empty on interim transcripts; final results carry the resolved language.

Concurrent requests
Shared active-session limits: Free 1, Pro 20, Growth 100, Enterprise negotiatedSource for Concurrent requests (opens in a new tab)

A prerecorded in-flight request or open realtime/Turns connection each consumes one slot. Limits are checked at request start; sessions in progress complete.

Billing basis
2 credits per audio minute; diarization adds 0.2 credits per minuteSource for Billing basis (opens in a new tab)

Free accounts stop accepting new requests once the credit limit is reached. Paid plans have no credit cap, but concurrency limits still apply.

Session duration
Not collected

No maximum session duration established by reviewed docs. A 30-second ping interval is a keepalive cadence, not a session limit.

Maximum audio file
Not collected

No numeric upload ceiling established by the reviewed prerecorded contract.

Provisional live transcripts
Optional include_interim=true; default sends final transcripts onlySource for Provisional live transcripts (opens in a new tab)

Persist final text after is_final=true. Later interim text can revise previous partials.

Utterance endpointing
flush_request finalizes buffered audio, followed by flush_confirmationSource for Utterance endpointing (opens in a new tab)

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 13 Oct 2026
CategoryUsageRateUnit
Speech recognitionPro realtime overage1€0.012audio minute
Speech recognition add-onPro diarization add-on overage2€0.0012audio minute
Speech recognitionGrowth realtime overage3€0.008audio minute
Speech recognition add-onGrowth diarization add-on overage4€0.0008audio minute

Rate conditions

  1. Pro costs €20/month including 6,000 shared credits. This overage rate applies only after included credits: 2 credits/minute × €0.006/credit. It is not a standalone pay-as-you-go offer or an all-in average. No USD conversion or model-release pinning is implied. Exact rounding and tax treatment require the account terms.
  2. Pro overage: 0.2 extra credits/minute × €0.006/credit. Additive to base transcription after shared included credits are exhausted. Requires the corresponding paid subscription; no diarization on Turns.
  3. Growth costs €100/month including 30,000 shared credits. This overage rate applies only after included credits: 2 credits/minute × €0.004/credit. It is not a standalone pay-as-you-go offer or an all-in average. No USD conversion or model-release pinning is implied. Exact rounding and tax treatment require the account terms.
  4. Growth overage: 0.2 extra credits/minute × €0.004/credit. Additive to base transcription after shared included credits are exhausted. Requires the corresponding paid subscription; no diarization on Turns.
Pricing source (opens in a new tab)

Data & enterprise terms

Hosted audio is processed on Reson8-owned infrastructure in the Netherlands, with no audio subprocessors according to the security guide. Audio is not retained after processing unless explicitly agreed otherwise. Terms retain terminology lists for the agreement and delete them within 30 days after termination, subject to legal exceptions. Audio and terminology are not used for general-model training without separate informed consent. Zero audio retention does not mean no stored account or terminology data; the website privacy policy has a different scope.

Read the service policy (opens in a new tab)

Reson8 API · turns

Provider source (opens in a new tab)
API model ID
Not collected

API path selects a task/transport; custom_model_id selects a terminology adapter, not an immutable base-model version.

Access
Connect to wss://api.reson8.dev/v1/speech-to-text/turns. Server clients use Authorization: ApiKey or Bearer. Browser clients use a short-lived bearer token in Sec-WebSocket-Protocol. Send binary audio and receive JSON text events.
Regions
Netherlands · provider-owned GPU infrastructure
Languages
German (de), English (en), Spanish (es), French (fr), Frisian (fy), Italian (it), Dutch (nl), Polish (pl), Portuguese (pt), Swedish (sv)Source for Languages (opens in a new tab)
Streaming
SupportedSource for Streaming (opens in a new tab)
Batch processing
Not supportedSource for Batch processing (opens in a new tab)
Audio formats
WAV, M4A, MP3, OGG, FLAC, WebM, Raw PCM s16le, G.711 μ-law, G.711 A-lawSource for Audio formats (opens in a new tab)

Container auto-detection is default. Prerecorded also accepts explicit seekable media containers; fresh streaming connections need headers or explicit raw encoding.

Audio channels
1–10 input channels; default 1Source for Audio channels (opens in a new tab)

Transport channel count is distinct from max_speakers, whose maximum is eight on diarization endpoints.

Audio sample rates
Default 16,000 Hz when used with explicit encoding; exact full range unverifiedSource for Audio sample rates (opens in a new tab)

Recommended raw streaming input is 16 kHz mono PCM s16le.

Custom vocabulary
Up to 250 per-request phrases or 50,000 phrases per named custom model; no stated limit on number of custom modelsSource for Custom vocabulary (opens in a new tab)

Query custom_model_id overrides the API client default. bias_strength defaults 0.45 and is non-negative. Larger bias can introduce irrelevant terms.

Research controls
Patterns recover expected structured identifiers: up to 32 expanded characters excluding separators and 32 alternative shapesSource for Research controls (opens in a new tab)

Patterns cannot combine with phrases, custom_model_id, or an API-client default custom model. They may worsen transcription when the expected token is absent.

API request fields
language, phrases or custom_model_id, bias_strength, include_timestamps, include_words, include_language, include_confidence, patterns and filler_modeSource for API request fields (opens in a new tab)

Not all combinations are valid. Filler mode is clean, natural (default) or verbatim; realtime interim text is unaffected.

Speaker diarization
Not supportedSource for Speaker diarization (opens in a new tab)

Turn segmentation is not multi-speaker diarization. Homepage FAQ says all endpoints; the dedicated guide explicitly excludes Turns.

Word timestamps
Optional include_words and include_timestamps; include_confidence adds word confidenceSource for Word timestamps (opens in a new tab)

These fields occur on turn_end_candidate; turn_end confirms the latest candidate.

Concurrent requests
Shared active-session limits: Free 1, Pro 20, Growth 100, Enterprise negotiatedSource for Concurrent requests (opens in a new tab)

A prerecorded in-flight request or open realtime/Turns connection each consumes one slot. Limits are checked at request start; sessions in progress complete.

Billing basis
2 credits per audio minuteSource for Billing basis (opens in a new tab)

Free accounts stop accepting new requests once the credit limit is reached. Paid plans have no credit cap, but concurrency limits still apply.

Session duration
Not collected

No maximum session duration established by reviewed docs. A 30-second ping interval is a keepalive cadence, not a session limit.

Maximum audio file
Not collected

No numeric upload ceiling established by the reviewed prerecorded contract.

Utterance endpointing
Acoustic and semantic turn detection: eager_turn_probability default 0.5; final_turn_probability default 0.92, each in [0,1]Source for Utterance endpointing (opens in a new tab)

turn_start supports barge-in; candidates replace earlier candidates and turn_end confirms the latest. These are probability thresholds, not silence durations.

Provisional live transcripts
turn_start → zero or more replacement turn_end_candidate events → turn_end confirmationSource for Provisional live transcripts (opens in a new tab)

A flush_request forces the current turn to end with a final candidate and confirmation.

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 13 Oct 2026
CategoryUsageRateUnit
Speech recognitionPro turns overage1€0.012audio minute
Speech recognitionGrowth turns overage2€0.008audio minute

Rate conditions

  1. Pro costs €20/month including 6,000 shared credits. This overage rate applies only after included credits: 2 credits/minute × €0.006/credit. It is not a standalone pay-as-you-go offer or an all-in average. No USD conversion or model-release pinning is implied. Exact rounding and tax treatment require the account terms.
  2. Growth costs €100/month including 30,000 shared credits. This overage rate applies only after included credits: 2 credits/minute × €0.004/credit. It is not a standalone pay-as-you-go offer or an all-in average. No USD conversion or model-release pinning is implied. Exact rounding and tax treatment require the account terms.
Pricing source (opens in a new tab)

Data & enterprise terms

Hosted audio is processed on Reson8-owned infrastructure in the Netherlands, with no audio subprocessors according to the security guide. Audio is not retained after processing unless explicitly agreed otherwise. Terms retain terminology lists for the agreement and delete them within 30 days after termination, subject to legal exceptions. Audio and terminology are not used for general-model training without separate informed consent. Zero audio retention does not mean no stored account or terminology data; the website privacy policy has a different scope.

Read the service policy (opens in a new tab)

Reson8 · private deployment

Provider source (opens in a new tab)
API model ID
Not collected

API path selects a task/transport; custom_model_id selects a terminology adapter, not an immutable base-model version.

Access
Contact Reson8 for licensed on-premises or VPC deployment. The Obsidian customer story documents Helm installation on AWS EKS with prerecorded transcription and diarization; no self-service container download or immutable public model selector is published.
Regions
Customer-selected VPC/on-premises environment, subject to written agreement
Deployment targets
Customer-managed AWS EKS via Helm is documented; Enterprise also advertises on-prem/VPCSource for Deployment targets (opens in a new tab)
Batch processing
SupportedSource for Batch processing (opens in a new tab)
Speaker diarization
SupportedSource for Speaker diarization (opens in a new tab)
Billing basis
Not collected

Private deployment is negotiated; hosted subscription rates are not copied to this endpoint.

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

Data & enterprise terms

Provider documents a customer deployment through Helm on AWS EKS. Residency, logging, retention, capacity, network and upgrade policies depend on that customer environment and its agreement; hosted Netherlands guarantees are not automatically inherited.

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.

Reson8 API · prerecorded

Not verified

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

How it works
Not collected

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

Minimum reusable input
Not collected

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

Lifetime & refresh
Not collected

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

How to enable it
Not collected

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

Usage fields
Not collected

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

Billing rules
Not collected

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

Limits & data boundaries
Not collected

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

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

Reson8 API · realtime

Not verified

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

How it works
Not collected

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

Minimum reusable input
Not collected

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

Lifetime & refresh
Not collected

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

How to enable it
Not collected

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

Usage fields
Not collected

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

Billing rules
Not collected

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

Limits & data boundaries
Not collected

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

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

Reson8 API · turns

Not verified

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

How it works
Not collected

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

Minimum reusable input
Not collected

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

Lifetime & refresh
Not collected

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

How to enable it
Not collected

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

Usage fields
Not collected

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

Billing rules
Not collected

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

Limits & data boundaries
Not collected

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

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

Reson8 · private deployment

Not verified

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

How it works
Not collected

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

Minimum reusable input
Not collected

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

Lifetime & refresh
Not collected

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

How to enable it
Not collected

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

Usage fields
Not collected

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

Billing rules
Not collected

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

Limits & data boundaries
Not collected

No discounted prompt-cache or model-cache API verified. Speech buffering and stored terminology adapters are not prompt caching.

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

The terms retain rights in base models, adapter architecture and inference engines and license use of a proprietary hosted speech system. Custom terminology adapters cannot be exported. This record explicitly represents the current managed model family, not another independently named weight release.

Classification evidence (opens in a new tab)

Sources reviewed