Model Releases
How Full-Stack NIM Optimizations Deliver 2.5x More Users on Nemotron 3 Ultra
NVIDIA’s Nemotron 3 Ultra NIM 2.0.12 stack boosts a 4×B200 system’s throughput by up to 2.5 ×, achieving roughly 1,997 tokens per second at a target of 50 transactions‑per‑user through autotuned kerne
NVIDIA’s Nemotron 3 Ultra NIM 2.0.12 stack boosts a 4×B200 system’s throughput by up to 2.5 ×, achieving roughly 1,997 tokens per second at a target of 50 transactions‑per‑user through autotuned kernels, tensor parallelism, prefix/state reuse, scheduler/memory tuning, and MTP speculative decoding. The NIM is delivered as a deployable microservice bundle with validated GPU‑aware configurations, standard APIs, and enterprise‑ready container lifecycle; developers can benchmark it against their own traffic using NVIDIA AIPerf to meet desired latency SLOs.
Related
- NVIDIA Nemotron 3 Ultra Leads Open Models on Accuracy and Efficiency in Agentic RTL Coding
- NVIDIA Nemotron 3 Ultra Powers Faster, More Efficient Reasoning for Long-Running Agents
Source: NVIDIA Developer | 2026-09-10