Amazon launches new G7e instances for generative AI on SageMaker.

With the launch of G7e instances based on NVIDIA RTX PRO 6000, Amazon SageMaker offers developers powerful tools for generative AI.

Amazon Web Services (AWS) has announced the launch of new G7e instances that accelerate the processing of generative AI using Amazon SageMaker services. These instances are equipped with NVIDIA RTX PRO 6000 Blackwell GPUs and provide access to 96 GB of GDDR7 memory per GPU. The new G7e allows developers to work with powerful models such as GPT-OSS-120B and Qwen3.

G7e instances are available in configurations with 1, 2, 4, and 8 GPUs and significantly outperform their G6e predecessors in performance. For example, G7e.2xlarge supports models with up to 35 billion parameters, while G7e.48xlarge handles models with up to 300 billion parameters. The new instances provide incredible data transfer speeds of up to 1,600 Gbps, which enhances multithreading.

With the launch of G7e, AWS strengthens its position in the cloud computing market by offering new solutions for large companies working with generative AI. G7e instances provide up to 768 GB of total memory and increase request processing speed by 2.3 times compared to G6e, which is an important competitive advantage.

The launch of G7e could fundamentally change the implementation of high-performance applications such as chatbots, text processing, and image generation. Increased memory and speed will help organizations reduce request processing times and simplify operations, leading to cost savings and increased productivity when working with new AI models.