AI inference costs dropped up to 10x on Nvidia's Blackwell — but hardware is only half the equation

New deployment data from four inference providers shows where the savings actually come from — and what teams should evaluate ...
Network World

Nvidia claims 10x cost savings with open-source inference models

Nvidia noted that cost per token went from 20 cents on the older Hopper platform to 10 cents on Blackwell. Moving to Blackwell’s native low-precision NVFP4 format further reduced the cost to just 5 ...
Neowin

NVIDIA announces TensorRT-LLM for Windows that boosts LLMs by up to 4 times with RTX GPUs0 0

NVIDIA has announced TensorRT-LLM for Windows. This open-source library will allow PC developers with NVIDIA GeForce RTX graphics cards to boost the performance of LLMs by up to four times. John ...
Datacenter Dynamics

Nvidia sets benchmarking performance records with its H200 and TensorRT-LLM software

Nvidia has set new MLPerf performance benchmarking records on its H200 Tensor Core GPU and TensorRT-LLM software. MLPerf Inference is a benchmarking suite that measures inference performance across ...
Search Engine Land

Bing Search gets faster, more accurate and efficient through SLM models and TensorRT-LLM

The Bing Search team shared how it helped make Bing Search and Bing’s Deep Search faster, more accurate and more cost-effective by transitioning to SLM models and the integration of TensorRT-LLM. Bing ...