Nvidia says its open-source AI model matches closed rivals at lower cost

Nvidia said its open-source Nemotron 3 Ultra matched leading closed AI models on LangChain's enterprise benchmark at 10x lower cost per run.blogs.nvidia+1
The 550-billion-parameter model, released in June, uses a hybrid architecture supporting up to one million tokens of context for complex agentic workflows.research.nvidia+1
Nvidia shares rose Wednesday after the announcement, with Box , Abridge, Amdocs, and EY among early enterprise adopters embedding the model.tradingview

Nvidia on Tuesday announced that its open-source Nemotron 3 Ultra AI model has achieved performance parity with leading closed AI systems on enterprise benchmarks — at one-tenth the inference cost — sending shares up roughly 4% on Wednesday as investors digested the competitive implications for the broader AI market.

A New Benchmark for Open Models

The announcement, made in partnership with LangChain, centered on results from LangChain's Deep Agents benchmark suite. According to Nvidia, Nemotron 3 Ultra achieved "business task parity with the highest-scoring closed models" while completing more tasks at higher throughput and running at 10 times lower inference cost per run. No model retraining was required; all performance gains came from engineering the environment around the model.blogs.nvidia+1

Nemotron 3 Ultra is a 550-billion-parameter Mixture-of-Experts model with 55 billion active parameters, built on a hybrid Transformer-Mamba architecture that supports context lengths up to one million tokens. Originally released on June 4, 2026, the model was designed for long-running agentic workflows including coding agents, deep research, and complex enterprise tasks.openrouter+3

"The way to build better agents is to keep improving the system around the model," said Harrison Chase, cofounder and CEO of LangChain, in a statement accompanying the announcement. "Our work with NVIDIA shows that enterprises can get strong performance from an open stack while keeping control over the agent systems they are building."blogs.nvidia

Enterprise Adoption and the NemoClaw Blueprint

Nvidia pointed to early enterprise adoption as evidence of the model's commercial readiness. Abridge, Amdocs, and Box are embedding specialized agents into their platforms, while EY is expanding its Nvidia implementation capabilities around the NemoClaw for LangChain Deep Agents blueprint. The blueprint packages the tuned model with Nvidia's OpenShell secure runtime, giving enterprises a fully open stack they can customize and deploy on their own infrastructure.tradingview+1

LangChain's agent engineering platform processes more than 200 million monthly downloads, providing broad distribution for the tuned Nemotron 3 Ultra profile. Developers can access the model through Baseten, Crusoe Cloud, DeepInfra, Fireworks, Nebius, and Together AI.blogs.nvidia

Market Response

Nvidia shares gained approximately 4% on Wednesday following the announcement, according to TradingView, as investors responded to the model's competitive positioning against proprietary alternatives. The rally came alongside a Bank of America reiteration of its Buy rating on the stock on July 8. Nvidia's stock has gained about 5% year-to-date in 2026, trailing the broader market after a period of consolidation.finance.yahoo+2

Sources (25)