nova-3

nova-3 is a advanced-tier language model from Deepgram. nova-3 is a proprietary model available exclusively through Deepgram's API. The Nova family is Deepgram's model series for this capability tier.

nova-3 is priced at $0.218 per hour. This is a speech-to-text model billed per hour rather than per token — token input/output pricing does not apply.

Its 0-token context window is standard for this tier — adequate for most chat and completion tasks, but may require chunking for long-document workloads. Median latency (P50 time-to-first-token) is approximately 300ms — suitable for interactive applications where sub-second response is acceptable.

nova-3 is best suited to transcription, real-time, low-latency. Supported capabilities include streaming token output for real-time UIs. When evaluating this model for a specific workload, the most important factors are: context window (does your input fit?), output price (how long are your generations?), and latency (is this interactive or batch?). The pricing table above shows how nova-3 compares to similar models across these dimensions.

nova-3 sits in the advanced tier — strong general-purpose models balancing capability and cost, suitable for most production applications. Unlike open-weights alternatives, nova-3 is only available via Deepgram's API — there is no self-hosted option, which simplifies deployment but creates vendor dependency. When comparing nova-3 to alternatives, use the similar models section on this page to find models at comparable price points with different capability tradeoffs. The intelligence score (0/100) reflects aggregate benchmark performance — a useful proxy for general capability, though real-world performance varies significantly by task type. Always test with your specific prompts and use case before committing to a model for production.