Skip to content
View original post on X: merve
Published
38/100

NVIDIA releases Nemotron 3.5 Bolt E4B, an 8B model with a DFlash drafter

AINVIDIA has released Nemotron 3.5 Bolt E4B, an 8B-parameter model with 4B effective parameters, which Merve Noyan says outperforms Gemma-4 and Qwen3.5. The release includes a DFlash drafter for speeding up decoding, along with quantized versions from ggml and Unsloth.
merveVerified on X
@mervenoyann
View on X

NVIDIA just released an E4B version (8B, 4B effective) of Nemotron 3.5 Bolt that outperforms Gemma-4 and Qwen3.5 🤯

it even comes with DFlash drafter for speeding up decoding ⚡️

comes with quants from ggml & unsloth too! https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Bolt-E4B-BF16

Source details

merve · x.comPublished · added here

Original title: NVIDIA just released an E4B version (8B, 4B effective) of Nemotron 3.5 Bolt that outperforms Gemma-4 and Qwen3.5 🤯

Topics

NVIDIAOpen-source ecosystemChina AIModel releasesEvals & benchmarks

Tags

#Hugging Face