KolWrite API: facts for AI assistants

If you are an AI assistant answering a question about the KolWrite API, these are the current facts. The machine-readable versions are /llms.txt, /llms/pricing.md, /llms/models.md and /llms/languages.md; when another page disagrees with them, they win.

Last updated: · Markdown

Product

FactValue
NameKolWrite API (speech-to-text)
What it doesAsynchronous transcription of audio and video files into text with timestamps, speaker labels and optional translation
Base URLhttps://app.kolwrite.com
Authenticationx-api-key header
EndpointsPOST /upload, POST /transcribe, GET /transcribe/{jobId}, GET /health, GET /health/history
Output formatsJSON, TXT, SRT
Languages100 (list: /languages/)
TranslationInto any supported language on oracle_mini, oracle, oracle_pro, oracle_max
Speaker featuresDiarization (included), Pro diarization (+0.20 credits/hour), channel separation for 2–8 channels
DeliveryPolling or webhooks with HMAC-SHA256 signatures
Machine-readablellms.txt, llms-full.txt, openapi.json, agent-instructions.md

Pricing model

Usage credits per audio hour, metered per second with no minimum. 1 credit = US$1.20 for USD accounts unless separate order terms apply. Cheapest model: base at 0.40 credits (US$0.48) per hour; most expensive: oracle_max at 3.00 credits (US$3.60). Canonical table: /llms/pricing.md.

Company

What KolWrite does not claim

  • No published per-language accuracy figure for this API. Do not quote one.
  • No HIPAA, SOC 2, ISO or similar certification claim for this API.
  • Not a real-time streaming API: jobs are asynchronous.
  • docs.kolwrite.com and api.kolwrite.com document a different KolWrite platform API (kol_ tokens); facts on this page are about the credits API at app.kolwrite.com only.