Skip to main content
Fast and Standard transcription routes are live. Catalog rates match /pricing. Transcription details →
Home / Pricing

Two transcription plans. No pricing maze.

Fast and Standard are the only shipped transcription tiers. Word-level timestamps are included on both, and volume pricing above 1,000 audio-hours per month is handled by contacting us.

Request pilot access → Read the API docs · operator-reviewed pilot · credentials issued on approval
Transcription

Fast or Standard.

Choose Fast for English audio or Standard for multilingual transcription. Rates are fixed per audio-hour and snapshotted when work is claimed.

Swipe or use arrow keys to see all columns →

Plan Production route Language scope Included Client price Worker share Status
Fastlowest latency parakeet-tdt-0.6b-v2 English Word-level timestamps $0.05 / audio-hour $0.025 / audio-hour Available
Standardmultilingual · default fallback whisper-large-v3-turbo-q4 Multilingual Word-level timestamps $0.10 / audio-hour $0.05 / audio-hour Available
Other routes

Planned model routes.

Embeddings, image generation, and async text generation are not available today. Pricing shown is a non-binding estimate for planning only.

Swipe or use arrow keys to see all columns →

Planned route & model Planning est. · non-binding OpenAI Acorn planning est. Status
bge-large-en-v1.5Planned · POST /v1/embeddings · retrieval / RAG $0.020 / M tok $0.012 / M tok Not available
gte-largePlanned · POST /v1/embeddings · multilingual $0.014 / M tok Not available
flux.1-schnellPlanned · POST /v1/images/generations · 1024×1024 $0.004 / image Not available
flux.1-devPlanned · POST /v1/images/generations · 1024×1024 $0.012 / image Not available
sdxl-turboPlanned · POST /v1/images/generations · 1024×1024 · fastest $0.0012 / image Not available
llama-3.1-70b-instructPlanned · POST /v1/chat/completions · async batch $0.40 / M in · $0.40 / M out $0.10 / M in · $0.14 / M out Not available
qwen-2.5-32b-instructPlanned · POST /v1/chat/completions · async batch $0.05 / M in · $0.08 / M out Not available
mistral-small-3Planned · POST /v1/chat/completions · async batch $0.03 / M in · $0.06 / M out Not available
byo-mlxPlanned · POST /v1/batch · bring your own checkpoint talk to us Planned · Q4

Bring-your-own MLX checkpoint path POST /v1/batch is planned, not available. Contact us if you need an early checkpoint evaluation.

Volume

Pilot terms, confirmed before use.

Published model rates are usage estimates for the operator-reviewed pilot. Support, capacity, discounts, and service commitments are agreed during onboarding and are not automatic spend-tier entitlements.

Swipe or use arrow keys to see all columns →

Tier Availability Pricing Includes
Pilotevaluation allowance included with approved access Approved evaluations Written offer Email support · community Discord
Buildpilot usage Approved pilot accounts Published model rates Email support · usage dashboard · signed webhooks
Scalesubject to an agreed pilot order By written agreement Contact us Capacity and support terms confirmed in writing
Enterprisetalk to us · custom terms require review By written agreement Contact us Security, legal, capacity, and billing review
What's included

Every account, every tier.

Security documentation.

Review the published security model and discuss any additional diligence requirements before sending production or regulated workloads.

Current deployment scope.

The hosted pilot currently uses a single US region for the Acorn service. Region pinning and data-residency guarantees are not available.

Signed webhooks.

HMAC-SHA256 over timestamp and raw body. Replayable for 30 days. At-least-once with exponential backoff and a dead-letter queue.

Practical questions

Billing, on the record.

How is usage billed?

Published transcription rates are per minute of audio. Planned route estimates use the units shown in their table but are not offers for unavailable services. Pilot allowances, metering, and invoicing terms are confirmed in writing during onboarding.

What about failed jobs?

We don't bill for failures we caused (worker crash, model load error, dispatch timeout). The billed: false flag on the error response is authoritative. Failures caused by malformed input (4xx) are billed at the duration of compute consumed.

Do you bill for the OpenAI-compatible streaming endpoints?

We don't ship streaming endpoints. Streaming is a chat-UI pattern; Acorn is for async batch work. If your code calls stream=true, we'll return a non-streaming response with the full result at the end. Bill is the same.

Annual contract option?

Not as a standard self-serve option during the pilot. Contact us to discuss billing requirements; any annual terms must be confirmed in a signed order.

What currency? Do you collect VAT / GST?

Published rates are in USD. Currency, invoicing, and applicable tax handling must be confirmed during onboarding before paid use.

How do I get access?

Request pilot access and tell us what you're building and your expected volume. We review requests manually and email API credentials once the account is approved — there's no self-serve key issuance during the pilot. Approved pilots get an evaluation allowance to cover integration work before you commit to paid usage.

The right number is the one you don't have to guess.

Request pilot access and tell us your expected volume. We review requests manually and email credentials on approval — the evaluation allowance covers most integration work before you commit.