Deepgram
#5 in Automatic Speech Recognition · Rated 4.5
Deepgram: fast, accurate STT with flexible pricing tiers
Deepgram offers high-speed speech recognition with multiple model families, including standard, enhanced, and Whisper-based options. The Pay-As-You-Go tier provides a low-cost baseline, with Growth and Enterprise plans offering prepaid credits and additional discounts, including high-concurrency and self-host capabilities. The platform emphasizes real-time streaming with low latency, broad language support, and a unified API for STT, TTS, and audio intelligence. In 2025, Deepgram differentiates itself with Whisper-based options and specialized agent capabilities, making it attractive for call centers, media workflows, and large-scale voice-enabled apps. The pricing model, which bills per minute, plus a self-serve API, makes it a strong fit for developers who want predictable costs at scale.
Features
- Real-time streaming and pre-recorded transcription
- Whisper-based models and multilingual support
- Audio intelligence (topic detection, summarization, etc.)
- Self-hosted enterprise option
- Text-to-Speech and Voice Agent APIs
Pros
- Competitive per-minute rates with multiple model options
- Fast real-time transcription with low latency
- Self-hosted deployment for enterprise needs
Cons
- Pricing complexity across models and plans
- Smaller ecosystem compared with big cloud providers