AWS adds over 100 models from 30 providers to JumpStart in 2026

The 2026 expansion of SageMaker JumpStart reflects a deliberate effort to broaden the model catalog beyond a concentrated set of providers. The body notes that the catalog now includes 30 providers, up from a more concentrated set in prior years. This breadth is positioned as a way to give customers more options for performance, cost, licensing, and language support, and to reduce concentration risk in model selection.

The onboarding priorities themselves signal where AWS sees customer demand. The body groups the new models into several themes: expanding the NVIDIA open-model portfolio with the Cosmos3 world foundation models for physical AI applications, closing the multimodal gap with models that natively handle vision, speech, and document understanding, making creative AI accessible for image and video generation, prioritizing efficient and quantized models for cost-sensitive production workloads, onboarding purpose-built models for targeted tasks like OCR and reranking, and broadening provider diversity with regional contributors such as BharatGenAI and Kakao.

Beyond model additions, the 2026 enhancements—the simplified Model Hub on the SageMaker AI console, pre-configured deployment options, and the Hugging Face deep link integration—are described as reducing friction across the model lifecycle. The body frames these changes as lowering the barrier between model discovery and production deployment, with the new Model Hub surfacing the full catalog on the console landing page and the deep link integration skipping manual setup steps like IAM role configuration. The combined effect, according to the body, is to help customers build production AI applications on AWS with less manual configuration.