Researchers have introduced Realtime-Venus, a novel full-duplex interaction system designed for more natural human-computer communication. The system utilizes two specialized 9B models: Realtime-Venus-Omni for audio-visual tasks and Realtime-Venus-Audio for spoken interactions. These models integrate continuous perception, conversational control, and speech generation through a shared timeline, allowing for asynchronous background reasoning and tool execution without interrupting foreground dialogue. Realtime-Venus-Omni demonstrates strong performance on video benchmarks, while Realtime-Venus-Audio leads in audio understanding and spoken question answering, outperforming models like Gemini 3.1 Live and GPT-4o in continuation metrics. AI
IMPACT Sets a new standard for interactive AI systems, potentially influencing future multimodal and conversational agent development.
RANK_REASON Paper release detailing a new AI system and its performance on benchmarks.
Read on Hugging Face Daily Papers →
- Daily-Omni
- Full-Duplex-Bench v1.5
- Gemini 3.1 Live
- GPT-4o
- Llama Questions
- MMAU-Pro
- OVO-Bench
- Realtime-Venus
- Realtime-Venus-Audio
- Realtime-Venus-Harness
- Realtime-Venus-Omni
- Speech CMMLU
- StreamingBench
- VoiceBench AlpacaEval
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →