NVIDIANemotron 3.5
Nemotron 3.5 Bolt E4B
Bolt E4B is NVIDIA's further post-training of Google's Gemma 4 E4B instruction model. This release accepts and returns text.
What the release provides
It has 8B total and 4B effective parameters. The card lists input and output limits up to 64K. BF16 weights support customization, while a separate 4GB version targets limited-memory deployment.Nemotron 3.5 Bolt E4B official model card
How to use it well
Practical guidance. Do not infer image or audio input from the capabilities of the Gemma parent. Connect speech recognition and synthesis as separate components when building a voice interface. Review the exact quantized version on the target device before selecting it.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Model identifiers | nvidia/NVIDIA-Nemotron-3.5-Bolt-E4B-BF16 |
| Release license | OpenMDW 1.1; Gemma parent under Apache 2.0Nemotron 3.5 Bolt E4B official model card |