← All models

Nemotron 3.5 Bolt E4B (precision unspecified)

NVIDIA · 2026-10-23

NVIDIA Nemotron 3.5 Bolt E4B labels whose evaluators do not identify the served BF16 or 4GB artifact. NVIDIA published Bolt E4B on October 23, 2026 as a small model further post-trained from Gemma 4 E4B-it for multi-turn tool use, instruction following and input grounding in local voice and device assistants: a BF16 checkpoint (nvidia/NVIDIA-Nemotron-3.5-Bolt-E4B-BF16, 8B total and 4B effective parameters) and a 4GB build quantized to Q4_K_M with llama.cpp importance-matrix calibration. NVIDIA reports its benchmark table on BF16 and its charts on Q4_K_M. Bolt E4B takes text input, has a 64K-token context and optional thinking with a runtime budget, under OpenMDW-1.1. This benchmark entry preserves the source observations without assigning them to a checkpoint or including the unresolved identity in the capability fit.

Benchmark scores