Ollama adds NVIDIA’s compact Nemotron 3.5 Bolt E4B model

Ollama now offers NVIDIA’s Nemotron 3.5 Bolt E4B, a compact model designed for conversations and tool-based tasks on local computers. Its listing includes support for Windows, Linux and Apple Silicon.

Ollama describes E4B as having 8 billion parameters, with 4 billion effective, and says NVIDIA developed it through additional training of Gemma 4 E4B. The model supports instruction following and tasks involving several exchanges with tools. A companion draft model can accelerate generation through speculative decoding.

E4B joins a larger Bolt variant in NVIDIA’s Nemotron 3.5 family. NVIDIA positions the smaller model for applications including local voice assistants, interactive characters and embedded interfaces, and had scheduled its availability for October 23. Developers can now select the E4B variant in Ollama for local deployment.