Skip to main content
Every transcription request must specify a model. Typeless offers three tiers that trade off quality and speed against cost — pick the one that fits your workload.

Models

ModelPriceQualitySpeedBest for
typeless-asr-l1-v1$0.25 / audio hourGoodSlowerCost-sensitive, high-volume batch workloads
typeless-asr-l2-v1$0.50 / audio hourBetterFastRecommended default — balanced quality, speed, and cost
typeless-asr-l3-v1$2.50 / audio hourBestFastestQuality-critical or latency-sensitive production workloads
  • l1 — The budget tier: lowest cost, but slower to return results and slightly lower transcript quality. Pick it when cost matters more than turnaround.
  • l2 — Sits between l1 and l3 on quality, speed, and price; the right default for most applications.
  • l3 — The top tier: highest accuracy and lowest latency, powered by the full multi-model pipeline. Pick it when transcript quality or response time is critical.

Billing rules

  • Billed per second of audio, with a 15-second minimum per request
  • You are not charged for validation failures (4xx), server-side errors, or timeouts
  • You are charged if you cancel or disconnect after transcription has started — billed on the audio you sent

Example

A 42.5-second file on typeless-asr-l2-v1:
cost = max(42.5, 15) s × ($0.50 / 3600 s) = 42.5 × $0.0001389 ≈ $0.0059
The minimum kicks in only for very short clips (under 15 seconds), so this file is billed at its actual duration.

Balance

Typeless uses a prepaid balance model. Top up your account in the Typeless Console. When your balance reaches $0, requests return 402 INSUFFICIENT_BALANCE. See Errors for handling guidance.