PRICING
Pay for processing. Nothing else.
Prepaid credits, drawn down per processed second. Pay for the stages a request actually runs, not seats, not bundles.
01
01 · RATES
Per processed minute.
STAGE · MODELBATCHSTREAMING
Speech-to-textULTRAFIELD ASR-1PRO$0.0021/min · $0.13/hr$0.0031/min · $0.19/hr
Speech-to-textWHISPER LARGE-V3-TURBO$0.0012/min · $0.07/hr$0.0018/min · $0.11/hr
Speech-to-textWHISPER LARGE-V3$0.0018/min · $0.11/hr·
Diarization$0.0010/min · $0.06/hr·
Speaker ID$0.0008/min · $0.05/hr·
BILLED PER SECOND · STAGES PRICED INDEPENDENTLY
02
02 · PLANS
Prepaid credits, three ways to run.
Top up credits; usage draws them down. No postpaid metering.
01
Pay as you go
Published rates
ONE-TIME TOP-UPS
+One-time top-ups; no subscription, pay only for what you run
+Every model and processor at the published per-minute rates
+Full API surface: batch + streaming, pipelines, webhooks, SSE
+Up to 10 concurrent batch jobs and 5 concurrent streams
+Standard rate limits
02
FEATUREDPro
$49/mo
OR $39/MO BILLED ANNUALLY
+Everything in Pay as you go
+$49 of usage included every month
+10% discount on all usage above the included allowance
+Pro ASR models
+Up to 50 concurrent batch jobs and 25 concurrent streams
+Priority job scheduling
+Spend controls and usage alerts
+Live chat support
03
Enterprise
Custom
COMMITMENTS + SLAS
+Volume discounts
+EU data center
+Custom concurrency, throughput, and rate limits
+Uptime and support SLAs
+Pay by invoice, security review, and DPA
+Model customizations and dedicated hosting
+Priority support and integration help
+Capacity planning for burst and sustained volume
Every plan uses the same API and the same per-second metering. We never train on customer data.
03
03 · FAQ
Pricing questions.
What shapes usage?
Model choice, audio duration, the processing stages a request runs, and LLM tokens. Every response reports its own usage.
How should we plan for volume?
Rates stay flat as you scale; volume rates are available for sustained workloads on Pro and Enterprise.
Do you train on customer data?
No. We never train on customer data.
Get started
Give your product better ears.
Move from raw audio to accurate transcripts, speakers, entities, sentiment, and application-ready output through one pipeline.
Batch and streaming · Proprietary and open models