What happened?
NVIDIA announced that the Nemotron 3.5 Bolt E4B, an open language model of about 4 GB optimized for conversational agents and AI characters, is scheduled to become available on October 23, 2026. It is part of the Nemotron 3.5 family expansion, which includes the 120-billion-parameter Super VL and the 24 GB Bolt 74B-A9B, already released on October 15.
According to NVIDIA’s official blog, the Bolt E4B combines natural conversation with multi-turn agentic capabilities to power fast local voice assistants, interactive characters and embedded interfaces. It is designed to run on GeForce RTX, Jetson and RTX/DGX Spark systems.
Why does this matter?
The Nemotron 3.5 Bolt E4B represents an advance in running artificial intelligence locally on consumer hardware, especially relevant for games and interactive entertainment. Smaller, efficient models allow developers to integrate smarter NPCs, dynamic dialogues and game agents without relying exclusively on the cloud, reducing latency and costs.
The Nemotron family is open, with weights and datasets available, making customization easier for studios and developers. Super VL is already used by companies such as Atlassian, CrowdStrike, Perplexity and Salesforce in agentic workflows, while the Bolt models target the edge and desktop.
What changes in practice?
With the arrival of the Bolt E4B, game creators and interactive application developers can experiment with AI characters that run locally on PCs and devices with NVIDIA GPUs, offering more natural and contextual responses without sending data to external servers. This opens the way for more immersive experiences in RPGs, narrative adventures and creation tools.
The model was still scheduled for October 23 at the time of the October 15 announcement. Developers can follow availability on Hugging Face and NVIDIA’s official channels. More information on the official NVIDIA blog.
Sources: Official NVIDIA blog, VentureBeat reporting and other coverage of the Nemotron 3.5 family launch.
Image credit: NVIDIA — Source: official NVIDIA blog
By GeekikiBot
