Description
Compact reasoning model from NVIDIA for on-device use, post-trained from Gemma 4 E4B with supervised fine-tuning and multi-environment reinforcement learning.
Model type
Language model
Stats
Capabilities
Minimum system memory
Tags
Last updated
Updated 2 days agobyREADME
NVIDIA Nemotron 3.5 Bolt E4B
Compact reasoning model from NVIDIA for on-device use, post-trained from Gemma 4 E4B with supervised fine-tuning and multi-environment reinforcement learning. Uses the E4B architecture's Per-Layer Embeddings for about 4B effective parameters.
Accepts text input. Features a reasoning toggle to enable or disable intermediate reasoning traces, and a conversational mode for brief, spoken-style replies. Trained for tool use, instruction following, multi-turn conversation and structured output. Supports English, Spanish, French, German, Italian, and Japanese.
Supports a context length of 128K tokens.
Custom Fields
Special features defined by the model author
Enable Thinking
: boolean
(default=true)
Controls whether the model will think before replying
Conversational Mode
: boolean
(default=false)
Makes replies brief and lightly formatted, for spoken or casual conversation
Parameters
Custom configuration options included with this model
Sources
The underlying model files this model uses