Skip to content
Model infoModelsSnapshot: 15 Oct 2026

Fish Audio / Transcribe

Fish Audio Transcribe 1 Pro

Speech recognition for recordings up to one hour, with optional word timestamps, speaker turns and vocal-event annotations.

availableClosed weightsSpeech-to-text
Official documentation (opens in a new tab)Reviewed 13 Oct 2026
11 of 22 specification fields collected
Audio durationUp to 60 minutes; clips below about 0.08 seconds are rejectedSee specification notes below
Audio formats7 optionsSee specification notes below
Input billingSee ratesFish Audio API
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
Reasoning
Not collected
Tool calling
Not collected
Structured output
Not collected
Streaming
Not supportedSource for Streaming (opens in a new tab)

Synchronous file transcription returns after the whole file is processed. Async SDK calls do not establish streaming or a batch service.

Batch processing
Not collected
Fine-tuning
Not collected
Knowledge cutoff
Not collected
Parameters
Not collected
Architecture
Not collected
Languages
Automatic multilingual detection with an optional ISO language hint; complete language inventory uncollectedSource for Languages (opens in a new tab)
Context caching
Not collected
Output formats
JSON transcript, Optional word-level segments, Optional speaker turnsSource for Output formats (opens in a new tab)
Published price
$0.36 per audio hourSource for Published price (opens in a new tab)
Audio duration limits
Up to 60 minutes; clips below about 0.08 seconds are rejectedSource for Audio duration limits (opens in a new tab)
Speaker attribution
Inline speaker markers; timestamped speaker_turns unless diarize=false; best-effort speaker-count hintsSource for Speaker attribution (opens in a new tab)

Speaker IDs are stable within one response, not across requests. Word segments do not expose per-segment speaker IDs.

Audio controls
Optional emotion and vocal-event tags via tag_audio_events; diarize and speaker-count hintsSource for Audio controls (opens in a new tab)
Audio formats
WAV, MP3, AAC/M4A/MP4, FLAC, Ogg Opus/Vorbis, WebM/Matroska, MOVSource for Audio formats (opens in a new tab)

Video containers use the first audio track. Raw PCM, AIFF, CAF, WMA, AMR and AC3 are unsupported.

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
transcribe-1-pro
Access
Native API reference labels this endpoint beta. Send the exact model header to avoid silently selecting Transcribe 1.
Regions
Processing region and residency not established for this access path.
API endpoint URL
https://api.fish.audio/v1/asrSource for API endpoint URL (opens in a new tab)
Request quotas
Shared account concurrency: 5 below $100 total prepaid; 15 from $100; 50 from $1,000; enterprise customSource for Request quotas (opens in a new tab)

Thresholds measure total prepaid funds, not spend. All keys and ASR share the pool; 429 responses have no Retry-After header. Free-tier fair-use conditions also apply.

Streaming
Not supportedSource for Streaming (opens in a new tab)

Synchronous file transcription returns after the whole file is processed. Async SDK calls do not establish streaming or a batch service.

Required request fields
One audio file via multipart/form-data or application/msgpack. Base64 audio in native JSON is unsupported. Model belongs in the HTTP header, not the body.Source for Required request fields (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 recognitionTranscription1$0.36audio hour

Rate conditions

  1. Full recording including silence, rounded up to the next second, once per successful request. Error responses are not billed. One request holds a shared concurrency slot until complete.
Pricing source (opens in a new tab)

Data & enterprise terms

Fish Audio terms permit Usage Data and Content to improve its models, including third-party components. Public terms do not establish universal zero retention. Free S2.1 has no DPA or TTFA guarantees; confirm production contractual terms separately. Voice inputs require appropriate rights.

Read the service policy (opens in a new tab)
API model ID
fish-audio/transcribe-1-pro
Access
Listed by OpenRouter. Upstream availability, account access and routing policy apply; host listing date is not the native model release date.
Regions
Processing region and residency not established for this access path.
API endpoint URL
https://openrouter.ai/api/v1/audio/transcriptionsSource for API endpoint URL (opens in a new tab)
Required request fields
Base64 JSON input_audio or OpenAI-compatible multipart upload; select model in the request bodySource for Required request fields (opens in a new tab)
Maximum audio file
25 MB inline/multipart cap for these providers; larger-body exceptions are documented for OpenAI/Groq, not Fish AudioSource for Maximum audio file (opens in a new tab)
Request mode
Complete JSON transcript; host guide warns of a 60-second upstream processing timeoutSource for Request mode (opens in a new tab)

This is processing time, not maximum recording duration. Native Pro’s one-hour contract does not waive host upload/time limits.

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 13 Oct 2026
CategoryUsageRateUnit
Speech recognitionTranscription1$0.0001audio second

Rate conditions

  1. OpenRouter published host rate. Rates are not transferred to other hosts.
Pricing source (opens in a new tab)

Data & enterprise terms

OpenRouter prompt/response logging and improvement use are opt-in; metadata is retained. Upstream provider policies remain separate, so routing through OpenRouter does not itself establish Fish Audio zero retention.

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.

Fish Audio API

transcribe-1-pro
Not verified

No model-specific prompt-cache contract or discount established. Voice references and previous-chunk conditioning are not billed prompt caching.

How it works
Not collected

No model-specific prompt-cache contract or discount established. Voice references and previous-chunk conditioning are not billed prompt caching.

Minimum reusable input
Not collected

No model-specific prompt-cache contract or discount established. Voice references and previous-chunk conditioning are not billed prompt caching.

Lifetime & refresh
Not collected

No model-specific prompt-cache contract or discount established. Voice references and previous-chunk conditioning are not billed prompt caching.

How to enable it
Not collected

No model-specific prompt-cache contract or discount established. Voice references and previous-chunk conditioning are not billed prompt caching.

Usage fields
Not collected

No model-specific prompt-cache contract or discount established. Voice references and previous-chunk conditioning are not billed prompt caching.

Billing rules
Not collected

No model-specific prompt-cache contract or discount established. Voice references and previous-chunk conditioning are not billed prompt caching.

Limits & data boundaries
Not collected

No model-specific prompt-cache contract or discount established. Voice references and previous-chunk conditioning are not billed prompt caching.

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

OpenRouter

fish-audio/transcribe-1-pro
Not verified

No model-specific prompt-cache contract or discount established. Voice references and previous-chunk conditioning are not billed prompt caching.

How it works
Not collected

No model-specific prompt-cache contract or discount established. Voice references and previous-chunk conditioning are not billed prompt caching.

Minimum reusable input
Not collected

No model-specific prompt-cache contract or discount established. Voice references and previous-chunk conditioning are not billed prompt caching.

Lifetime & refresh
Not collected

No model-specific prompt-cache contract or discount established. Voice references and previous-chunk conditioning are not billed prompt caching.

How to enable it
Not collected

No model-specific prompt-cache contract or discount established. Voice references and previous-chunk conditioning are not billed prompt caching.

Usage fields
Not collected

No model-specific prompt-cache contract or discount established. Voice references and previous-chunk conditioning are not billed prompt caching.

Billing rules
Not collected

No model-specific prompt-cache contract or discount established. Voice references and previous-chunk conditioning are not billed prompt caching.

Limits & data boundaries
Not collected

No model-specific prompt-cache contract or discount established. Voice references and previous-chunk conditioning are not billed 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 official API documents this named managed model under Fish Audio service terms that reserve service IP and restrict reverse engineering. The reviewed official weight inventory does not publish this release; the separately downloadable S2 Pro and S1-mini are excluded. Exact internal checkpoint lineage remains unverified.

Classification evidence (opens in a new tab)

Lifecycle dates

Native model release18 Sept 2026Source (opens in a new tab)

Sources reviewed