SAVA Models by Inworld
Inworld releases · 3 in the last 12 months · this family litChangelog →
1 model2026
Last refreshed 2026-11-19. Next refresh: weekly.
Links
WebsiteAbout
Inworld’s acoustic speech-activity and end-of-turn model.
Decision facts
- Best fit
- voice activity detectioncoding
- Capability starting point
- SAVA
- Lowest tracked input
- Not tracked
- Closest related family
- Voice Isolator
Current Variants
Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.
1 in view
SAVACurrent
Use when the workload needs voice activity detection, 25.5M parameters, and audio.
2026-11voice activity detection25.5M parametersaudio
| Model | Use when | Released | Signals | Status |
|---|---|---|---|---|
| SAVA | Use when the workload needs voice activity detection, 25.5M parameters, and audio. | 2026-11 | voice activity detection25.5M parametersaudio | Current |
Release Timeline
1 release group2026-11
1 current
SAVA
Currentvoice activity detection25.5M parametersaudio
Specifications(1 models)
| Model | Released | Parameters |
|---|---|---|
| SAVA | 2026-11 | 25.5M |