The whole dayThu 15 Oct 2026
AI Compass
ArchiveAbout

ModelNVIDIAConfirmed

Story 3 of 10 in the Thursday 15 October edition

NVIDIA releases Nemotron 3.5 Super VL and Bolt open models for agents

Super VL, a 120B model that also reads images and documents, targets always-on enterprise agents, while Bolt fits a 74B model into 24GB of memory for RTX PCs and DGX Spark.

Edited by Ashutosh Shrivastava
16 Oct 2026, 10:47 IST · 2 min read

Five dark tiles with gold icons for search, code, charts and video around a glowing green cube, in front of Earth seen from space at night
Image: NVIDIANVIDIA’s header image for the announcement

NVIDIA released Nemotron 3.5 Super VL and Nemotron 3.5 Bolt 74B-A9B on 15 October 2026. Super VL has 120B parameters, 12B of them active per token, and a 1M-token context window. It reads text, images and video, with a focus on what agents meet at work: web pages, tables, charts and PDFs.

NVIDIA says Super VL is comparable in accuracy with its largest reasoning model at about one-fifth of the parameters, and five times faster than the next-best open model in its class.

Nemotron 3.5 Super VL sits alone in the chart’s fast, high-scoring corner, at about five times the output speed of Inkling Small, which has a similar Intelligence Index score
Artificial Analysis Intelligence Index against output speed for open-weight models, as published by NVIDIAImage: NVIDIA

Bolt 74B-A9B is compressed from Super VL and stored at 2-bit precision, so it fits in 24GB and can run beside other apps on RTX PCs, DGX Spark and NVIDIA’s RTX Spark laptops, which arrive on 16 October. NVIDIA ranks it the top open model in its class on PinchBench. Bolt E4B, which needs 4GB, follows on 23 October for voice assistants and devices such as Jetson.

Super VL comes with open weights, open datasets and training recipes, and both released models use the OpenMDW-1.1 licence. Super VL is on Hugging Face, OpenRouter and ModelScope and as an NVIDIA NIM microservice, and companies including Scale AI, Ramp and CrowdStrike are customising it with their own data. NVIDIA’s announcement lists no prices.

After partners’ post-training, Super VL gains 9% to 44% on five specialist tasks and scores above the closed model shown for each, such as Claude Opus 5 on medical insurance claims
NVIDIA’s chart of accuracy gains from partners’ post-training on five agent tasksImage: NVIDIA

Sources

  1. PrimaryOpen From Data Center to Desktop: NVIDIA Expands Nemotron Family With Highest Efficiency Super VL and Bolt Modelsblogs.nvidia.com · NVIDIA blog, 15 October 2026
  2. OfficialNVIDIA-Nemotron-3.5-Super-VL-120B-A12B-NVFP4 model cardhuggingface.co/nvidia · Parameters, context window, inputs and licence
  3. OfficialNVIDIA-Nemotron-3.5-Bolt-74B-A9B-24GB model cardhuggingface.co/nvidia · Bolt’s 2-bit build, inputs and licence
  4. On X@nvidia on X: expanding the Nemotron family with Nemotron 3.5 Super VLx.com · NVIDIA’s launch post

The rest of the day

Read the whole day

Spotted an error? Email aiforsuccess.in+compass@gmail.com or tag @ai_for_success on X