Nvidia launches Nemotron 3.5 Super VL, a 120B-parameter open model for enterprise agents, and says it outperforms frontier models on specialized benchmarks
VentureBeatCarl Franzen
Context & Ripple Effects
Nvidia has been building Nemotron as a range rather than a single model: the initial Nemotron 3 family covered 30B, 100B and roughly 500B sizes, followed by a 120B Nemotron 3 Super release in March 2026 and a 30B multimodal Nano Omni variant in April.
The August Lightning model and NeMo Switchyard routing library shifted the family toward agent deployment. Super VL extends that arc with a larger open multimodal option, while Nvidia's claimed specialized-benchmark lead gives enterprise evaluators a vendor-defined comparison point rather than an independently established result.
First-order effects
Enterprise agent teams gain a 120B open multimodal Nemotron option, and vLLM's announced day-zero support lowers the delay between release and serving evaluation.
Nvidia expands the set of agent workloads it can address through Nemotron, with its benchmark claims placing Super VL directly into enterprise model-selection tests.
Second-order effects
NeMo Switchyard gains a higher-capacity Nemotron target alongside the 30B Lightning model, giving agent operators a basis to route different tasks across Nvidia's own open-model lineup.
Nvidia's smaller Lightning and larger Super VL releases make model size and workload fit a more explicit buying choice for enterprises evaluating open agent stacks.
Third-order effects
If enterprises adopt the lineup, Nvidia's sequence of base, multimodal, routing and size-specific releases points to open weights becoming a portfolio strategy for agent infrastructure rather than a one-model release cycle.
The trend: Nvidia is assembling an open Nemotron stack in which model scale, multimodality and routing are tailored to enterprise agent deployment.
Meet NVIDIA Nemotron 3.5 Super VL Our new open model combines reasoning and tool use with text, image, document, and video understanding for long-running AI agents. 120B total parameters with 12B active 🧵
@nvidia
@nvidia on x
Today, we're expanding our Nemotron family of open models with Nemotron 3.5 Super VL. Bringing visual understanding, reasoning, and tool use together to help AI agents complete more work in less time. Open models ➡️ more control to customize, deploy, and build on your own terms…
@artificialanlys
@artificialanlys on x
Nvidia's Nemotron 3.5 Super VL is now the leading US open weight model, scoring 26 on the Artificial Analysis Intelligence Index Nemotron 3.5 Super VL achieves twice the intelligence of its predecessor, Nemotron 3 Super (13), and does so using ~60% fewer output tokens per task It…
@eliebakouch
Elie on x
multiple release by nvidia today, love how they compress nemotron 3.5 super to the “perfect” size for their specific hardware (RTX spark). there is a full tech report, including how they use their own speculative decoding method that they found better than DSpark
@nvidiartxspark
@nvidiartxspark on x
NVIDIA Nemotron 3.5 Bolt 24GB puts powerful local agents within reach. ⚡ 74B-A9B parameters packed into 24GB of memory, balancing intelligence and speed while running alongside your apps. Optimized for RTX Spark and ready for your next project.
@nousresearch
@nousresearch on x
Hermes Agent is the best place to use the new @NVIDIAAI Nemotron 3.5 Super VL and Bolt models. NVIDIA benchmarked Super VL on Hermes Bench with a score of 63.44, and compared to similar local models it has up to 5x the token throughput.
@tinkerapi
Tinker on x
Nemotron 3.5 Super VL from @nvidia is live on Tinker today. With 12B active parameters, efficient architecture, and native text, images, and video inputs 3.5 Super VL should be a strong base for training long running always-on multimodal agents.