Speed tier
4 text-to-speech (TTS) models, ordered speed first: by speed tier and P95 time to first audio on Coval, then the models Coval does not measure, by name. No combined score.
Data checked on October 22, 2026 · updated after each benchmark capture
4 text-to-speech (TTS) models are in the Headroom speed tier: P95 at or under 101 ms on Coval's 30-day window, measured from US East. Headroom to spend elsewhere in the turn.
Coval calls each hosted API every 30 minutes and publishes percentiles over 30 days. The tiers group models on the P95 time to first audio, the slowest one turn in twenty, and break where the data has gaps. Within a tier, latencies within 10 ms are a tie. Voice quality, accuracy and price are compared on the model pages, never added into one score.
Alibaba Qwen3 TTS Fast is an open-weights text-to-speech (TTS) API from Alibaba (Qwen), served by Nari Labs, measured by two independent benchmarks: Coval and Speko.
Fluxions vui is a proprietary text-to-speech (TTS) API from Fluxions, measured by one independent benchmark, Coval.
Gradium TTS is a proprietary text-to-speech (TTS) API from Gradium, measured by three independent benchmarks: Artificial Analysis, Coval and Speko.
Inworld Realtime TTS-2 Flash is a proprietary text-to-speech (TTS) API from Inworld, measured by three independent benchmarks: Artificial Analysis, Coval and Speko.