GPT Transcribe improves on its predecessor but can't catch ElevenLabs, Google, or Mistral on error rates

OpenAI launched two new speech-to-text API models: GPT Transcribe for prerecorded audio and GPT Live Transcribe for low-latency streaming.

GPT Transcribe is about 34x faster than real time, cuts word error rate to 3.31%, and lowers pricing by 25% to $0.0045 per minute.

Both models support text context, keywords, and multiple input languages, aiming to improve transcription quality and flexibility.

OpenAI is cheaper and more accurate than before, but it still trails ElevenLabs, Google, and Mistral on the AA-WER benchmark.