In the news
NVIDIA Nemotron 3.5 Transcribe and Granary v2: Intelligent Multilingual Transcription You Can Prompt and Adapt
Hugging Face · Published · 3 min read · 114 on Hacker News
In 30 seconds
- What happened
- NVIDIA released Nemotron 3.5 Transcribe, an open-weight model handling transcription, speaker attribution, translation, and audio Q&A across 40 languages.
- Why it matters
- Engineers building multilingual speech systems, medical dictation tools, or meeting transcription features need models that handle speaker labels and domain-specific vocabulary.
- Watch out
- Model is designed for recorded audio, not live streaming. Audio question answering trains only in English. Results require spot-checking against source recordings.
- prompt
- transcription
- nemotron
The patterns behind this
- MAPS: Multilingual Agent Performance & Security
- Agentic Context Engineering (Evolving Playbook)
- Error Handling and Recovery Patterns
Each one covers how the technique works, when it earns its cost, and where it breaks.
The Agent Architect
One pattern, one tradeoff, one production failure story. A short weekly briefing for people building agentic systems.
Weekly email, one-click unsubscribe. We only use your address to send the briefing.