Nemotron 3.5 Super VL (AI model): news timeline & CVEs

[AINews] Agents disprove Polysemanticity; greatly improve Interpretability & Alignment

Meta released Muse Spark Open, an 816B MoE that leads US open-weight models on Artificial Analysis.

Meta released Muse Spark Open, an 816B-total / 42B-active mixture-of-experts model with a 1M-token context and text, image, video, and audio input. Artificial Analysis scores it 39, first among US open-weight models but behind GLM-5.3 (45) and Kimi K3 (44); launch pricing is $1.25/$4.25 per million input/output tokens. The same roundup covers a Goodfire interpretability essay and launches of NVIDIA Nemotron 3.5 Super VL and Microsoft’s roughly 3T-parameter MAI-Base-2.

Latent SpaceOct 16, 04:16 UTC · 21h agoModel release 7 sources2· 1 read

NVIDIA Releases Nemotron 3.5 Super VL, a 120B Open Multimodal Agent Model

NVIDIA released Nemotron 3.5 Super VL, a 120-billion-parameter open multimodal agent model activating 12 billion parameters.

NVIDIA released Nemotron 3.5 Super VL, an open 120B mixture-of-experts vision-language agent model that activates 12B parameters per token and accepts text, images, documents, and short video. It offers up to a 1 million-token context, OpenMDW-1.1 weights, and NVFP4 and BF16 checkpoints runnable on hardware such as one H200, two H100s, DGX Spark, or one GB200. NVIDIA-reported NVFP4 scores include 89.90 on GPQA-Diamond, 71.07 on SWE-bench Verified, and 77.5 on BrowseComp, while tau3-bench banking scored 17.78. The model was distilled from Nemotron 3 Ultra after training on about 26 trillion tokens; NVIDIA also released Bolt edge models plus datasets and NeMo Gym recipes.

MarkTechPostOct 15, 19:53 UTC · 1d agoModel release 4 sources4· 1 read

Nvidia's Nemotron 3.5 Super VL is an open model built to run enterprise AI agents nonstop

Nvidia released open Nemotron 3.5 Super VL, a 120B multimodal model for always-on enterprise agents.

Nvidia released Nemotron 3.5 Super VL, an open 120-billion-parameter Mixture-of-Experts multimodal model with about 12 billion parameters active per token, for continuous enterprise agents handling text, images, documents, and video. It was distilled from Nemotron 3 Ultra with the Nemotron Coalition, including Mistral AI, Thinking Machines Lab, Perplexity, and Black Forest Labs. Nvidia says it matches Ultra on agent work at roughly one-fifth the size, beats Nemotron 3 Super on SWE-bench and Terminal-Bench, and scores 80.8 percent on CharXiv. With Dynamo on 16 B200 GPUs, Nvidia claims eight times as many concurrent users and an 85 percent lower time to first token. Bolt 74B-A9B targets 24GB local machines, and a 4GB Bolt E4B is planned later.

The Decoderupdated Oct 15, 19:53 UTC · 1d agofirst Oct 15, 16:09 UTC · 1d agoModel release 4 sources3· 2 reads

Open From Data Center to Desktop: NVIDIA Expands Nemotron Family With Highest Efficiency Super VL and Bolt Models

NVIDIA released open Nemotron 3.5 Super VL, a 120B agent model, plus Bolt edge models.

NVIDIA released Nemotron 3.5 Super VL, an open 120-billion-parameter vision-language model for always-on multimodal agents, claiming accuracy comparable to its largest reasoning model at about one-fifth the size and up to 5x faster throughput than peers. Bolt 74B-A9B, available now, is a 24GB 2-bit model for local agent tasks on RTX and DGX Spark, while Bolt E4B 4GB for conversational agents on GeForce RTX, Jetson, and Spark is due October 23. NVIDIA says organizations including Scale AI, CrowdStrike, Atlassian, Perplexity, and Salesforce are customizing Super VL. The release includes open weights, datasets, and training recipes.

NVIDIA Blogupdated Oct 16, 15:00 UTC · 10h agofirst Oct 15, 15:00 UTC · 1d agoModel release 5 sources3

Nemotron 3.5 Super VL

Timeline4 articles

Appears with

[AINews] Agents disprove Polysemanticity; greatly improve Interpretability & Alignment

NVIDIA Releases Nemotron 3.5 Super VL, a 120B Open Multimodal Agent Model

Nvidia's Nemotron 3.5 Super VL is an open model built to run enterprise AI agents nonstop

Open From Data Center to Desktop: NVIDIA Expands Nemotron Family With Highest Efficiency Super VL and Bolt Models