Long-horizon memory on JoyAI Video-Call Interaction Human Evaluation
77.8Win RateJoyAI-VL-Interaction
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| JoyAI-VL-InteractionComparison baseline=Gemini, Evaluation setting=In-app video-call assistant, Rating scale=Three-level (good, fair, poor)2026.06 | 77.8 | 22.2 | 0 | |
| JoyAI-VL-Interaction2026.06 | 77.8 | 22.2 | 0 | |
| DoubaoPlatform=in-app video-call assistant2026.06 | 0 | 22.2 | 77.8 |