Release dates
— NVIDIA released an open-weight model for local conversational agents with 4GB GGUF and BF16 checkpoints and up to 64K context. ↗
Nemotron 3.5 Bolt E4B
NVIDIA’s open-weight small conversational agent model with 8B total and 4B active parameters, targeting local voice assistants, interactive characters, and tool use.
MODEL NUMBER267
- First public date
- 2026-10-23
- Organization
- NVIDIA
- Family
- Nemotron
01
Specifications
Only published specifications are shown.
02
Overview
Why it mattered
NVIDIA released Nemotron 3.5 Bolt E4B on 2026-10-23, with 4GB GGUF and BF16 checkpoints and up to 64K input and output tokens. NVIDIA says it builds on Gemma 4 E4B with additional post-training; the archive has no matching E4B base entity, so no lineage edge is inferred.
03
Architecture & capabilities
Model capabilities
ReasoningTool useConversationLocal inferenceMultilingual
04
Release history
NVIDIA released an open-weight model for local conversational agents with 4GB GGUF and BF16 checkpoints and up to 64K context.
View source05
Related models
Select a card to view the related model.
No related models recorded yet
06
Sources
01
Introducing NVIDIA Nemotron 3.5 Bolt E4B 4GBNVIDIA Nemotron · Published 2026-10-23 · Checked 2026-10-24
Official · Verified02NVIDIA Nemotron 3.5 Bolt E4B model card and weightsNVIDIA (official Hugging Face organization) · Checked 2026-10-24
Official · Verified03NVIDIA Nemotron 3.5 Bolt E4B BF16 weightsNVIDIA (official Hugging Face organization) · Checked 2026-10-24
Official · VerifiedSuggest a correction
Copy this template, add the field, proposed value, and primary source, and share it with the archive maintainer.