This product sheet outlines Deepgram’s enterprise speech models built for accuracy, speed, and scalability. Base models support multiple languages and dialects including English (US, GB, IN, NZ), Spanish, Korean, French, Portuguese, Russian, and Turkish. Designed for use cases such as general transcription, phone calls, meetings, and earnings calls, these models serve as the foundation for custom training. With sufficient labeled audio, organizations can tailor models to their domain. Deepgram’s deep learning approach allows rapid expansion into new languages while maintaining strong real-time transcription performance.

Join for free to read