NVIDIA Pairs Open Reasoning Models with NIM for Faster AI Agent Inference
NVIDIA announces open reasoning models that run up to 9x faster, available alongside OpenAI gpt-oss models through NVIDIA NIM.
NVIDIA is positioning its inference platform around open reasoning models designed for enterprise AI agents, according to the company's AI page. The models are built for planning and multi-step reasoning tasks and run up to 9x faster than previous approaches, targeting reduced costs for agent deployments across customer service, cybersecurity, manufacturing, logistics, and robotics.
The company is also making OpenAI gpt-oss models available through NVIDIA NIM, its inference microservices layer, allowing enterprises to deploy them on any GPU-accelerated infrastructure while maintaining data privacy. The combination of open reasoning models with NIM positions NVIDIA to capture demand from organizations building agentic AI systems that need fast, flexible inference without sending data to third-party APIs. For enterprise buyers evaluating agent platforms, the availability of both NVIDIA and OpenAI models through a single deployment stack simplifies infrastructure decisions.