Bring your own voice agent
A video-only avatar API turns incoming speech into synchronized avatar output. Your application supplies the conversation, voice generation, and any retrieval or tools.
Compare published video-only quotes
Sequencer Anima
Custom animation & moderation · Anima Fast / Max
HeyGen
Avatar-only integration · LiveAvatar
Beyond Presence
Enterprise deployments · Genesis 2.0 / A2V
Runway
Interactive characters · Characters / GWM-1
LemonSlice
Character styles & actions · LemonSlice-2.1 / CWM-1
Simli
Focused speech-to-video API · Emotive Faces / Trinity
| Detail | Sequencer Anima | HeyGen | Beyond Presence | Runway | LemonSlice | Simli |
|---|---|---|---|---|---|---|
| Model | Anima Fast / Max | LiveAvatar | Genesis 2.0 / A2V | Characters / GWM-1 | LemonSlice-2.1 / CWM-1 | Emotive Faces / Trinity |
| Full avatar usage rate | $5.99/hr | $11.40/hr | €21.00/hr | $12.00/hr | Not published | Not published |
| Full billing basis | Fast complete avatar | Essential Full overage | Starter managed overage | Native hosted session + $0.02/start | Hosted total not verified | Voice-stack total not verified |
| Video-only usage rate | $5.31/hr | $5.70/hr | €10.50/hr | $12.00/hr | $13.20/hr | $0.60/hr |
| Video billing basis | Fast video-only | Essential Lite overage | Starter video overage | Video session + $0.02/start | Starter video overage | Current Trinity Hobby quote |
| Full response timing | <1.8 s · Response-latency claim, vendor reported | Not verified · Full conversation timing not established | ~1.0–1.2 s · Managed-agent response, vendor claim | Not published · Hosted conversation timing not published | 2.04 s · Flash mean, reference voice stack (Ultra) | Not verified · Full response depends on voice stack |
| Video timing | Not verified · Video-only timing not separately published | <300 ms · First frames, vendor claim | ~100 ms · Speech-to-video model | Not listed · Comparable response timing not published | 471 ms · Flash mean time to first byte (Ultra) | <300 ms · Platform speech-to-video estimate |
| Output resolution | 640px / HD · 640px on Fast; HD on Max | 1080p on request · Needs appropriate source footage; HD increases latency | 1080p · Published Full HD output | Not published · No verified specification in the cited sources | Not published · No verified specification in the cited sources | Not published · No verified specification in the cited sources |
| FPS | 24 FPS · Published delivery target | Not published · No verified specification in the cited sources | 35 FPS · Published Full HD stream | Not published · No verified specification in the cited sources | 20 FPS · LemonSlice-2.1 platform claim | Not published · No verified specification in the cited sources |
| Custom motions | Custom motions · Create signature actions for live conversations | Recorded gestures · Source footage controls movement; no dynamic gesture control | Planned · Custom gesture controls described as next-generation features | Not verified · Prompted character behavior; signature actions not established | Actions (Ultra) · CWM-1 action and emotion control; separate from Starter model | Not verified · Audio-driven face animation; signature actions not established |
| Moderation control | Uncensored mode* · Custom filters; core safety limits remain | Not verified · An unmoderated or configurable mode has not been verified | Not verified · An unmoderated or configurable mode has not been verified | Not verified · An unmoderated or configurable mode has not been verified | Not verified · An unmoderated or configurable mode has not been verified | Not verified · An unmoderated or configurable mode has not been verified |
| Deployment | Hosted API | Hosted API | Hosted API | Hosted API | Hosted API | Hosted API |
Evaluate the whole delivery path
- Check audio format, chunking, and buffering requirements.
- Confirm the transport and avatar preparation steps.
- Measure audio arrival to relevant video output using one declared clock.
- Include the external speech stack when projecting total cost.
- Test reconnects, interruptions, idle animation, and session limits.
Next steps
Create a video-only shortlist · Estimate a video-only workload