Review · Software & AI
Deepgram review
A speech-to-text API with Nova-3 for general transcription and Flux for real-time voice agents, billed per second.
By the DBotMaker research desk · Updated November 7, 2026
- Capabilities (40%)
- 9.2
- Ease of setup (30%)
- 9.0
- Value (30%)
- 8.8
Deepgram offers two main speech models. Nova-3 handles general transcription in 60+ languages with diarization, smart formatting and keyterm prompting, and Flux STT targets voice agents with turn detection and interruption handling. Streaming prices on pay as you go are $0.0077 per minute for Nova-3 monolingual or Flux English and $0.0092 for Nova-3 multilingual, which works out to $0.46 and $0.55 per hour; Deepgram says pre-recorded audio costs less. Billing is per second without block rounding, and smart formatting is included, while redaction costs $0.0020 per minute and keyterm prompting $0.0013. The $200 starting credit needs no card and does not expire. Deployment is the main differentiator: you can use Deepgram's cloud, route through regional endpoints, or self-host, and you can train a custom model on your own data. The Growth plan cuts rates by up to 20% but starts at $4,000 a year.
Who it is for
Teams that want to self-host, need industry-specific models, or want a large free credit to prototype.
Price
Pay as you go with $200 in free credit; Nova-3 monolingual streaming $0.0077 per minute, multilingual $0.0092; pre-recorded runs lower; Growth from $4,000 a year.
Nova-3 Monolingual $0.0077 / min
Quoted from the vendor pricing page on November 7, 2026.
Key features
- Nova-3 in 60+ languages with speaker diarization, smart formatting and keyterm prompting
- Flux STT with turn detection and interruption handling for voice agents
- Multilingual models that handle code-switching mid-utterance
- Custom model training on your own data and industry models such as Nova-3 Medical
- Cloud, regional endpoints or self-hosted deployment
Strengths and weaknesses
- Strength: $200 free credit, which Deepgram says covers roughly 43,000 Nova-3 minutes
- Strength: Billed per second with no block rounding
- Strength: Self-hosting option for data residency and compliance
- Weakness: Streaming costs about three times AssemblyAI's Universal-Streaming rate
- Weakness: Growth discounts need a $4,000 yearly prepayment
- Weakness: No single published latency figure for Flux or Nova-3
Key facts
- Free credit: $200, no card required, does not expire
- Languages: 60+
- Deployment: Deepgram cloud, regional endpoints or self-hosted
Where it ranks
Deepgram is No. 2 in our Best Speech-to-Text APIs verdict and No. 3 in Best Text-to-Speech APIs.
Alternatives we scored
- AssemblyAI9.1
- Speechmatics8.8
- ElevenLabs Scribe8.7
- Gladia8.5
Visit deepgram.com · Questions or corrections: editorial@dbotmaker.io