Access · read on October 27, 2026
Open weights text-to-speech models, ranked by speed
12 of the 96 text-to-speech (TTS) models in this index are filed under Open weights , led by Alibaba Qwen3 TTS Fast, Alibaba Qwen3 TTS 1.7b and Fish Audio Drama 3 Preview. Same order and same measures as the full ranking.
| Rank | Model | Speed tier | P95 | Speed (shorter is faster) | Median | Listener Elo | Price per 1M chars | Best for |
|---|---|---|---|---|---|---|---|---|
| 4 | | Headroom | 98.8 ms | 62.9 ms | Not rated | $10 | Voice agents | |
| 5 | | Fast enough | 136.6 ms | 103.6 ms | Not rated | Not collected | Voice agents | |
| 10 | | Fast enough | 209.3 ms | 146 ms | 1,245 | $15 | Customer service, Voice agents, Assistants, Entertainment, Knowledge sharing | |
| 26 | Guava Daytona TTSGuava | Visible delay | 584.8 ms | 450 ms | Not rated | Not collected | Not rated | |
| 37 (tie) | | No tier | Not timed | Not timed | Not timed | 1,227 | $34 | Customer service, Assistants, Entertainment, Knowledge sharing |
| 49 (tie) | | No tier | Not timed | Not timed | Not timed | 1,108 | $15 | Customer service, Assistants, Entertainment, Knowledge sharing |
| 52 (tie) | | No tier | Not timed | Not timed | Not timed | 1,088 | $16 | Customer service, Assistants, Entertainment, Knowledge sharing |
| 62 (tie) | | No tier | Not timed | Not timed | Not timed | 1,043 | $15 | Knowledge sharing |
| 63 (tie) | | No tier | Not timed | Not timed | Not timed | 1,048 | Not collected | Customer service, Assistants, Entertainment, Knowledge sharing |
| 63 (tie) | | No tier | Not timed | Not timed | Not timed | 1,045 | $15 | Customer service, Assistants, Entertainment, Knowledge sharing |
| 79 (tie) | | No tier | Not timed | Not timed | Not timed | 913 | $40.4 | Knowledge sharing |
| Unranked | | No tier | Not timed | Not timed | Not timed | Not rated | $5 | Not rated |
Frequently asked questions
- Which Open weights TTS model is the fastest?
- Alibaba Qwen3 TTS Fast, with a P95 time to first audio of 98.8 ms in Coval's 30-day window. Of the 12 models on this page, 4 are timed.
- How are these models ordered?
- As in the full ranking: speed tier, then P95 time to first audio, with ties within 10 ms ordered by median time to first audio. Models Coval does not time follow, ordered by listener Elo; models with neither measure come last, in alphabetical order.