Gemini 3.1 Flash Live: Google's latest audio model with improved benchmarks and watermarking
What's new in Gemini 3.1 Flash Live
Google has released Gemini 3.1 Flash Live, their highest-quality audio and voice model designed for real-time dialogue. The model delivers improved speed and natural rhythm for voice-first AI applications.
Key technical details
- Benchmark scores: 90.8% on ComplexFuncBench Audio (multi-step function calling with constraints) and 36.1% on Scale AI's Audio MultiChallenge (complex instruction following with "thinking" on)
- Improved capabilities: Better tonal understanding, recognition of acoustic nuances like pitch and pace, and dynamic adjustment to user frustration or confusion
- Watermarking: All audio generated includes SynthID watermark for AI content detection
- Multilingual support: Available in over 200 countries and territories
Availability and access
- For developers: Available in preview via Gemini Live API in Google AI Studio
- For enterprises: Included in Gemini Enterprise for Customer Experience
- For general users: Accessible via Search Live and Gemini Live
The model enables building voice-ready agents that handle complex tasks in noisy environments and supports longer conversation threads during extended interactions.
📖 Read the full source: HN AI Agents
👀 See Also
Autoresearch Pushes Qwen3.5-397B to 20.34 tok/s on M5 Max via SSD Streaming
A developer achieved 20.34 tokens/second inference speed for the 209GB Qwen3.5-397B model on a MacBook Pro M5 Max with 128GB RAM using SSD streaming and 36 systematic experiments. The result represents a 2x speedup over the M5 Max baseline and 4.67x over the original M3 Max result.
Opus 4.7 Can Follow ~500 Instructions, Up from ~150 a Year Ago
Research updated in May 2026 shows Opus 4.7 can reliably follow ~500 instructions, compared to ~150 in July 2025. GPT-5.5 handles ~5000. Implications for CLAUDE.md file size.
Gemma 4 Released: Four Model Sizes for Local AI Hosting
Google has released Gemma 4 with four model sizes optimized for different hardware, including edge devices, laptops, and GPUs. All models are multimodal with text and vision capabilities, and the smaller models support real-time audio.
Hollywood Creatives Are Training AI to Replace Them — and Getting Paid $12–$200/hr
Award-winning writers, directors, and producers are taking gig work to train AI models—teaching them screenplay writing, production scheduling, and more—amid a 35% production slump.