Together AI has launched speech-to-text APIs powered by Whisper V3 Large, claiming 15x faster transcription than OpenAI while maintaining accuracy. Key differentiators include support for audio files exceeding 1GB (vs OpenAI's 25MB cap), seamless processing of 30+ minute recordings without chunking, word-level timestamp alignment, 50+ language support, and pricing at $0.0015 per audio minute. The service offers dedicated GPU endpoints for sub-second processing and batch processing for high-volume workloads. A Python SDK and interactive playground are available for immediate testing.
Table of contents
Performance That Changes What You Can BuildProduction-Ready API DesignBuilding Toward Complete Voice InfrastructureAvailable Now2 Impressions