Amazon Web Services expands SageMaker JumpStart with new AI models

Amazon Web Services has expanded its Amazon SageMaker JumpStart model catalog by adding several new foundation models from NVIDIA, Qwen, and Z. ai. These additions provide customers with specialized capabilities in areas such as agentic engineering, visual grounding, and multimodal reasoning.

New models include Z. ai’s GLM-5. 2 FP8, which features a 1M-token context window for long-horizon engineering tasks, and GLM-OCR for advanced document understanding. NVIDIA has introduced Nemotron-Nano-12B-v2, designed for high inference throughput in enterprise applications, as well as LocateAnything-3B, which focuses on visual grounding and object localization.

Additionally, Qwen has integrated the Qwen-AgentWorld-35B-A3B, a language world model capable of simulating agent environments across seven interaction domains, and the Qwen3. 5-122B-A10B, which utilizes a hybrid architecture for efficient multimodal reasoning. These models are available for deployment via the SageMaker console or the SageMaker Python SDK.