Nemotron 3.5 Embed, Rerank, and Parse: what's new (October 21, 2026)
NVIDIA published open Nemotron 3.5 retrieval models: Embed 8B BF16, Embed 1B BF16, Embed 1B NVFP4, Rerank 8B BF16, and Parse at 631 million parameters. The post says they are commercially available.
Who it is for
Readers following a public model or product announcement. AIToolStack does not score this as its own tool.
NVIDIA on Hugging Face (opens in a new tab)Browse the nine jobs