· 3 items
Google DeepMind ↗
Introducing Gemini 3.8 Live and 3.8 Live Extended Thinking
- What
- Google DeepMind released Gemini 3.8 Live and 3.8 Live Extended Thinking.
- Why
- Provides advanced live interaction and extended reasoning capabilities for agents.
- Impact
- Directly upgrades voice agent reasoning loops.
arXiv ↗
JustFit: 200K-Token LLM Serving on a 24 GiB Laptop with Just-in-Time…
- What
- JustFit enables 200K-token LLM serving on a 24 GiB laptop via JIT state management.
- Why
- Makes running high-context models locally practical without extreme hardware limits.
- Impact
- Improves local coding and agent development setups.
arXiv ↗
ECHO: A Matched-Contrast Benchmark for Context-Sensitive Turn-Taking…
- What
- ECHO introduced a matched-contrast benchmark for turn-taking in full-duplex dialogue.
- Why
- Addresses the core challenge of distinguishing interruptions from backchannels in voice.
- Impact
- Provides a direct evaluation tool for voice agent responsiveness.