AI News Roundup: Nemotron 3.5, Meta's Muse Glimmer, and Harmony Voice Agents

Here is the AI news from the last day that matters for teams building agents and voice workflows. Every item links to its source.

Nvidia releases Nemotron 3.5 Super VL and Bolt edge models

On October 19, Nvidia released Nemotron 3.5 Super VL, an open 120-billion-parameter multimodal model for enterprise agent work such as document analysis, tool use, and instruction following. Alongside it came Nemotron 3.5 Bolt 74B-A9B, which fits in 24GB of memory for local agents, with a smaller 4GB Bolt E4B aimed at voice interfaces and conversational agents due on October 23. (TechCrunch, SiliconReport)

Meta publishes Muse Glimmer 30B for on-device agents

Meta introduced Muse Glimmer, an open-source 30B model built for agentic tasks on consumer hardware. It takes text and image inputs, handles multi-step reasoning and tool use, is quantized to about 4-bit for single-GPU inference, and is released under the Apache 2.0 license. (AI at Meta)

monday.com launches Harmony, a voice AI platform for sales

Also on October 19, monday.com launched Harmony, a voice AI agent platform for sales teams. It runs small, task-specific language models that respond in about 200 milliseconds, and the company reports contact rates of up to 70% as agents work through pipelines businesses already paid for. (monday.com)

The pattern is clear: faster, smaller models are making voice agents practical for sales, not just support. If you want an AI voice agent that calls your leads, qualifies them, and hands ready buyers to a human closer, see Ozkra or get in touch.