Deepgram Flux TTS Deepgram · text-to-speech (TTS) API

Real-time latency

On Coval's 30-day window ending October 1, 2026 (11,765 runs), Deepgram Flux TTS has a P50 (median) time-to-first-audio (TTFA) of 193.9 ms, 9th lowest of the 31 models Coval measures. Its P99, the slowest 1% of turns, is 550.3 ms (18th lowest of the 31 models Coval measures). Latency consistency: a standard deviation of 131 ms (16th lowest of the 31 models Coval measures). Mean time to first byte (TTFB) is 140.5 ms. Speko measures a P50 TTFA of 106 ms from US East (n=30), dated July 3, 2026: 8th lowest of the 29 models Speko measures.

Accuracy (word error rate)

Coval measures an accuracy (word error rate) of 4.9% over 11,745 samples on the same window, 15th lowest of the 31 models Coval measures. Speko rates its pronunciation robustness on numbers, dates and currency at 0.2 on a 0-to-1 scale (22nd highest of the 28 models Speko rates).

Voice quality

Speko rates its naturalness at 1,550 Elo from blind A/B votes (18th highest of the 41 models Speko rates).

Price

Speko lists a cost of about $45 per 1M characters.

Scoring note

Speko changed its pronunciation robustness between the captures of September 30, 2026 and October 1, 2026 without a new measurement date (0.41, then 0.2): the score uses the average, 0.305. The value shown is the latest.