TTS latency consistency ranked: Gradium holds tightest 13ms spread, Gemini Flash widest
ArtificialAnlys · x · 2026-11-17
- Artificial Analysis adds variability data to its Time to First Audio benchmark (request percentiles):
- Gradium TTS is the steadiest overall, with the middle 50% of requests within a 13 ms spread.
- Speechify Simba 3.2 and Inworld Realtime TTS-2 are the steadiest leading models at 23 ms and 32 ms.
- Gemini 3.1 Flash TTS spans 908–975 ms; Mistral Voxtral 3 TTS spans 455–576 ms.
- From the parent thread: Eleven v4 Turbo (258 ms) and Simba 3.5 sit on the quality-vs-latency Pareto frontier.
Related event: Artificial Analysis Launches TTS Time-to-First-Audio Benchmark(4 posts)→
More from Models
- Fable 5.2 aces a MazeBench level 'suspiciously' well — fueling test contamination jokes — aiamblichus · 2026-11-17
- OpenAI to give 200,000 scientists free access to frontier models — OpenAI · 2026-11-17
- Claude's Incognito mode still reads your Gmail — economists flag misleading UX — Afinetheorem · 2026-11-17
- Benchmark finds AI rewrites overuse 'though' 5x more than humans: 0.89 vs 0.18 per 1k words — rseroter · 2026-11-17
- Blind test: Hy4 Flash beats Claude Haiku 5.5 at website generation for most humans — gnukeith · 2026-11-17
- First DeepSeek V4.1 Flash results on Vera Rubin NVL72: up to 93.3x per-chip throughput vs GB300 — vllm_project · 2026-11-17