Amazon SageMaker JumpStart: 2026 year-to-date in review

Summary by forck

Amazon says SageMaker JumpStart onboarded more than 100 new models from 30 providers between January and September 2026, widening the catalog customers can deploy from its machine learning hub.

• The company attributes the expansion to faster open source releases and customer demand for choice across architectures, sizes and capabilities.
• Alibaba's Qwen accounts for 29 of the additions, spanning reasoning, mixture-of-experts, agentic, code, speech, embedding and reranking models, while NVIDIA contributes 17, including the Cosmos3 world foundation models and Nemotron variants.
• Mistral added 7 models, Google 6 Gemma 4 variants, IBM 4 Granite models and Cohere 4, with 33 more from providers such as MiniMax, Zhipu AI, DeepSeek, and .