Announcing MCP Toolbox Java SDK v1.0: Agentic data access for the enterprise
cloud.google.com·6d ago
Cloud & Infrastructure
The New Stack·27d ago·5 min read
OpenAI’s voice model doesn’t think. That’s the point.
TL;DR
OpenAI and Google have introduced new voice models that address latency issues in voice agents. OpenAI's GPT-Live-1 separates real-time conversation from backend reasoning, while Google's Gemini 3.8 Live keeps reasoning within the voice model.
✦ Why It Matters
Evaluate your voice application architecture to optimize for real-time interactions and task execution.
Key Takeaways
- Consider using Gemini for seamless voice interactions with integrated reasoning.
- Use GPT-Live-1 for more control over backend processing and orchestration.
- Evaluate cost implications based on your voice agent's architecture.
- Test your voice model under real-world conditions to assess performance.
How It Works
Gemini 3.8 Live Extended Thinking keeps all processes within a single session, allowing it to maintain conversation flow while executing tasks asynchronously. This design minimizes latency and enhances user experience by enabling follow-up questions during tool execution.