Tabnine Adds Support for NVIDIA Nemotron Models, Bringing Powerful New Reasoning Capabilities to Enterprise AI

At Tabnine, our mission has always been simple: bring developers the best AI tools to write better software, faster, without sacrificing control, security, or flexibility. Today, we’re excited to share another step forward in that journey: support for NVIDIA’s brand-new Nemotron reasoning models.

These new models, including Llama Nemotron Super v1.5 and Nemotron Nano 2, are purpose-built for agentic AI applications and optimized for enterprise-scale performance. Starting now, they are part of the growing gallery of model options available through Tabnine’s secure, private AI development platform.

What Makes Nemotron Different?

Reasoning is quickly becoming a key differentiator in enterprise AI. Whether you’re deploying intelligent agents to help with code reviews, documentation, support workflows, or internal tools, you need models that can:

• Follow complex instructions
• Explore broader reasoning paths
• Maintain high throughput across teams
• Keep inference costs under control

That is exactly what NVIDIA built Nemotron to do. These are fully open, commercially viable models that outperform others in their class and they are now available in Tabnine, ready for real-world deployment.

Built for the Enterprise, from Day One

Nemotron models are available as NVIDIA Inference Microservices (NIMs), which makes them plug-and-play for customers running in the cloud, on-prem, or even in air-gapped environments. With support for up to 250 concurrent users on a single H100 and cost-efficient token generation, these models are engineered to scale.

And because Tabnine supports private, self-hosted deployments, you can take advantage of Nemotron’s performance without giving up control over your infrastructure or your data.

Open Models, Continuous Optimization

In keeping with our values, Nemotron is fully open: open model weights, open datasets, open tooling. That gives our customers freedom to fine-tune, customize, and build on top of a transparent foundation.

You can also take advantage of NVIDIA’s NeMo platform, including the NeMo Agent Toolkit and NeMo Data Flywheel, to build, monitor, and optimize your agents across their full lifecycle.

Continuing Our Work with NVIDIA

This is not our first collaboration with NVIDIA. In 2024, we announced support for NVIDIA NIM to give our customers even more flexibility in how and where they deploy Tabnine. Today’s announcement builds on that foundation, bringing even more choice and performance to our platform.

As the Nemotron family continues to evolve, we expect even better performance and more opportunities for developers to build AI agents that are not just fast, but also smart.

We are proud to work alongside NVIDIA to make that future a reality for our customers.

Want to learn more? Explore Tabnine’s enterprise AI platform and see how we’re helping teams build secure, private, and intelligent software at scale.