Speech-to-Text
WER
0.0%4.0%8.0%12.0%
Finalize latency
lower is betterTime from the caller stopping to the finalized transcript, not the first partial — the agent cannot reply until it lands.
0ms300ms600ms900ms1.20s
p50 p90
Time to first token
lower is better0ms2.00s4.00s6.00s8.00s
p50 p90
Cost
lower is betterWhat a vendor charges to transcribe a thousand minutes of audio, at its published streaming rate.
$0.00$5.00$10.00$15.00$20.00
ModulateVelma 2Sonioxstt-rt-v5InworldRealtime STT-1OpenAIGPT-4o-mini TranscribexAIGrok STTElevenLabsScribe v2OpenAIGPT TranscribeDeepgramNova-3Smallest AIPulseAlibabaQwen3-ASROpenAIGPT-4o TranscribeDeepgramFluxElevenLabsScribe v2 RealtimeAssemblyAIUniversal-3.5 ProCartesiaInk-2Gradium ASRGladiaSolaria-1GoogleChirp 3OpenAIGPT Live Transcribe