NVIDIA

(Santa Clara, CA)

www.nvidia.com

Builders of the open Nemotron family of high-throughput models - flagship Nemotron 3 Ultra plus the new hybrid-Mamba Nemotron 3.5 Super VL (120B MoE with image and video understanding).

2 MODELS TRACKED2 OPEN WEIGHTSNEWEST: NEMOTRON 3.5 SUPER VL · 1W AGO

What is the latest AI model from NVIDIA?

NVIDIA's primary flagship model is Nemotron 3 Ultra 550B (550B MoE (55B Activated), 262,144 tokens context). 550B open model with a generous free tier; top-10 OpenRouter volume for high-throughput open inference and experimentation. Their latest release is Nemotron 3.5 Super VL (Open-Weights Agentic VLM, released 2026-10-15). How we define “latest”

Release Timeline

NEWEST FIRST

October 2026[1 release]

Nemotron 3.5 Super VL

120B OPEN HYBRIDOct 15, 2026

NVIDIA's new open agentic vision-language model: 120B/12B-active Mamba-2-MoE hybrid with a C-RADIO vision encoder for image, document, and video understanding, 1M context, native MTP speculative decoding, and full open weights (BF16 + NVFP4) under OpenMDW-1.1; Terminal-Bench 2.1 60.67%, SWE-bench Verified 71.1%, GPQA-Diamond 89.52%, at $0.15/$0.50 per 1M tokens.

Open-Weights Agentic VLMOPEN DATASHEET

June 2026[1 release]

Nemotron 3 Ultra 550B

550B OPEN GIANT• FlagshipJun 4, 2026

550B open model with a generous free tier; top-10 OpenRouter volume for high-throughput open inference and experimentation.

Open High-ThroughputOPEN DATASHEET