Google DeepMind released StreamVibe video intelligence agent
The story so far
Google DeepMind has released the StreamVibe video intelligence agent, a model based on Qwen3.5-9B and specifically trained to understand ambiguous requests and provide help at the right time. On ProactiveBench, StreamVibe has improved response timing and quality by 10.7% and increased proactive interaction rates by 12.3%. Tests show it outperforms other open-source models, but it still struggles to continuously track changes in long videos (>5 minutes), with performance dropping to 38.5% of the baseline.
AI-written from the reports · Updated 17 days ago
Coverage timeline
Follow the coverage to see every side of the story.
- The DecoderGoogle wants video AI to understand what people need as they work
Google DeepMind released StreamVibe video agent, a model based on Qwen3.5-9B and specifically trained to understand ambiguous requests and provide help when appropriate, improving response timing and quality by 10.7% on ProactiveBench and increasing proactive interaction rate by 12.3%. Tests show it outperforms other open-source models, but it still struggles to track changes in long videos (>5 minutes), with performance dropping to 38.5% of the benchmark.