NVIDIA has released Nemotron 3.5 Bolt E4B 4GB, an open small language model post-trained from Gemma 4 E4B that runs in roughly 4GB of VRAM on RTX, Jetson and DGX Spark hardware. It adds multi-turn tool calling, a 64k context window and support for 35 languages, with 8B total parameters (4B effective) under the OpenMDW licence. https://www.marktechpost.com/2026/10/23/nvidia-nemotron-3-5-bolt-e4b-4gb/ #AIagent #AI #GenAI #OnDeviceAI
MarkTechPostNVIDIA Releases Nemotron 3.5 Bolt E4B 4GB: An Open Small Language Model for Local Voice Agents, Game Characters and RobotsNVIDIA Nemotron 3.5 Bolt E4B is a small language model for local conversational agents, running in ~4GB VRAM.