Side by side

GPT Transcribe vs Whisper v3 Turbo on Groq

Up to 4 models, normalised onto the same cost basis and scored through the lens you pick.

Cost of talking
Per minute$0.0045Not conversational
Per hour$0.270
30-min interview$0.135
List price$0.0045/minPublishedPlan-basedEstimate
Recruitment fit
Overall87/10062/100
VerdictThe right transcription default for interviews, because interviews are recorded in noisy rooms on laptop mics.Fine for internal notes; not what you want under an interview a hiring decision rests on.
Best for
Interview recordingsScreening call notesTechnical role discussions
Internal recruiter notesQuick voice memos
Watch out for
  • On clean studio audio the Balanced tier is genuinely more accurate — do not blanket-apply Smart
  • Small benchmark samples: re-validate on your own audio before standardising
  • Slowest tail latency of the streaming options we measured
  • No speaker attribution
Capabilities
Modalitiesaudioaudio
Context window
VariantsSmart, Balanced, Live (streaming)Turbo
On HumanLikeCoworker, ClippyCoworker
HumanLike evals
Word error rate (clean audio)0.29Our eval
Word error rate (with additive noise)0.26Our eval
Median latency (9.2s clip)1390 msOur eval
Text still arriving after speaker stops (EN)1197 msOur eval

Cost per hour assumes 150 wpm and a 40% AI speaking share. Hover a figure for its full derivation.