...

Navan Expands Its Open Model Strategy With NVIDIA Nemotron 3.5 Super VL

NVIDIA Nemotron 3.5 Super VL

Navan has put the NVIDIA Nemotron 3.5 Super VL into production. The global AI-powered business travel and expense platform announced the news today. The model now powers several of Navan’s AI travel agents. Notably, Navan trained and deployed it just three weeks after receiving early access and starting its work with NVIDIA.

This deployment marks the latest step in Navan’s open model strategy. First, the company evaluates models for specific travel tasks. Next, it adapts them with Navan’s own travel expertise. Finally, it self-hosts them to maintain control over performance, cost, and reliability.

How Navan Tested NVIDIA Nemotron 3.5 Super VL Before Launch

Before the rollout, Navan’s team evaluated Nemotron 3.5 Super VL internally. The team compared it against its self-hosted production baseline and other models. Each test used consistent prompts, tools, travel scenarios, and criteria.

Navan then fine-tuned the model with the NVIDIA NeMo AutoModel VLM LoRA training recipe. As a result, Nemotron 3.5 Super VL delivered the strongest overall performance. It beat or tied every competing model on six of seven measures. In addition, it recorded the lowest latency among the models tested.

Today, NVIDIA Nemotron 3.5 Super VL runs in production and powers customer conversations. Moreover, it does so at a lower cost. Navan users, therefore, receive greater value from the upgrade. The results also show significant efficiency gains for the company.

“Building reliable agentic travel systems involves much more than choosing a model, but continuously iterating to improve and solve real-world problems,” “We’re grateful to NVIDIA for early access to Nemotron 3.5 Super VL, which shows how open models can help us move faster, maintain control, and deliver meaningful value without compromising quality.” said Ilan Twig, co-founder and CTO of Navan.

Twig’s comments reflect Navan’s broader approach to agentic travel systems. Reliability, according to the company, depends on more than model selection. Instead, it requires continuous iteration to solve real-world problems. Early access to Nemotron 3.5 Super VL also gave Navan a head start. Similarly, open models help the company move faster while keeping control.

Open and Closed Models Power Navan’s AI Travel Agents

Navan combines open and closed models to build AI travel agents. These agents understand what travelers want. They also follow policies and use the right tools to get the job done. NVIDIA Nemotron 3.5 Super VL

Overall, the announcement shows how open models can support production-grade AI travel agents. Navan continues to blend open and closed models in its platform. Meanwhile, its focus stays on delivering value without compromising quality.

Explore IT Tech News for the latest advancements in Information Technology & insightful updates from industry experts!

News Source: Businesswire.com