GLM-5.2 FP8, NVIDIA-Nemotron-Nano-12B-v2 and GLM-OCR models now available on Amazon SageMaker JumpStart
AWS customers can now access three new AI models via Amazon SageMaker JumpStart: GLM-5.2 FP8, NVIDIA Nemotron-Nano-12B-v2, and GLM-OCR, expanding AWS’s foundation model portfolio.
Z.ai’s GLM-5.2 FP8 model is now available on Amazon SageMaker JumpStart, offering enhanced long-horizon agentic engineering capabilities. The model introduces a 1 million-token context window for the first time, enabling it to manage full development workflows, including requirements analysis, coding, testing, and deployment, within a single task. This represents a significant improvement over its predecessor, GLM-5.1, in handling extended project-level contexts and executing complex engineering tasks reliably.
NVIDIA’s Nemotron-Nano-12B-v2 has been added to SageMaker JumpStart, designed for unified reasoning and non-reasoning tasks with high inference throughput. The model leverages a hybrid Mamba-2 and Transformer architecture and a 128K context length, generating reasoning traces before delivering final responses. With a compact 12 billion parameters, it achieves accuracy comparable to or better than leading open models while offering up to six times higher inference throughput, making it suitable for enterprise applications requiring both precision and efficiency.
Z.ai’s GLM-OCR model is now accessible through SageMaker JumpStart, providing advanced document understanding for complex materials. The 0.9 billion-parameter multimodal model processes scanned PDFs, handwritten notes, academic papers with formulas, multi-column tables, and multilingual text. It reconstructs document structures, tables, and formulas into clean Markdown, JSON, or LaTeX formats with low latency, supporting real-time services and edge devices for large-scale document processing and invoice extraction workflows.
Customers can deploy any of these three models on AWS infrastructure using Amazon SageMaker JumpStart with minimal setup. Models can be accessed via the SageMaker JumpStart model catalog in the SageMaker console or deployed programmatically using the SageMaker Python SDK. AWS provides documentation to guide users through the deployment and utilization of foundation models within SageMaker JumpStart.