AWS expands custom LLM tools to speed model building

At AWS re:Invent this week, Amazon Web Services signaled that the next phase of enterprise AI is less about picking a single “best” chatbot model and more about tailoring models to a company’s exact data, language, and use cases. 

Coming right after AWS announced Nova Forge, a service for training custom versions of its Nova AI models, the company introduced new features in Amazon Bedrock and Amazon SageMaker AI meant to make custom large language model work faster and less intimidating for developers. (Szkutak) (“AWS doubles down on custom LLMs”)

A headline shift is “serverless model customization” in SageMaker, which AWS says lets teams start building or adapting a model without first planning infrastructure or compute resources. Ankur Mehrotra, AWS’s general manager of AI platforms, described it as removing the upfront friction that often slows experimentation: instead of provisioning hardware and pipelines first, developers can focus on the data and the technique they want to use. Access comes in two ways: a self guided, point and click workflow for more traditional development teams, and an agent led experience where developers can prompt SageMaker in natural language, which is launching in preview. (Szkutak) (“AWS doubles down on custom LLMs”)

Mehrotra offered a concrete example: a healthcare organization that wants a model to better understand medical terminology could provide labeled data, select an approach, and let SageMaker fine tune the model. Importantly, this is not limited to AWS’s in house models. AWS says the feature can be used to customize Nova models as well as certain open source models with publicly available weights, including DeepSeek and Meta’s Llama. That matters for enterprises that want flexibility and pricing leverage, but still need a clear path to adapting models to their own domain language, brand tone, and internal knowledge. (Szkutak) (“AWS doubles down on custom LLMs”)

AWS also introduced Reinforcement Fine Tuning in Bedrock, aimed at teams that want a more guided route to improving model behavior. Instead of designing an entire fine tuning run manually, developers can choose either a reward function or a preset workflow, and Bedrock will automate the customization process from start to finish. Together, these announcements reinforce that AWS is betting heavily on frontier LLM development and, especially, model customization as a differentiator for enterprises that do not want the same outputs as competitors using the same baseline models. (Szkutak) (“AWS doubles down on custom LLMs”)

That positioning is also a competitive necessity. AWS has not yet captured the kind of enthusiastic enterprise preference that some rival models enjoy; a Menlo Ventures survey from July reported that enterprises strongly favored Anthropic, OpenAI, and Google’s Gemini over other options. AWS’s answer is to make “different” easier than “bigger”: Nova Forge, announced during AWS CEO Matt Garman’s keynote, offers custom Nova models built for enterprise customers for $100,000 per year, while SageMaker and Bedrock updates aim to put customization tools more directly into developers’ hands. If AWS can make custom models feel routine rather than rare, it may turn a perceived weakness in model mindshare into an advantage in enterprise specificity. (Szkutak) (“AWS doubles down on custom LLMs”)

This article was written by the Covertly.AI team. Covertly.AI is a secure, anonymous AI chat that protects your privacy. Connect to advanced AI models without tracking, logging, or exposure of your data. Whether you’re an individual who values privacy or a business seeking enterprise-grade data protection, Covertly.AI helps you stay secure and anonymous when using AI. With Covertly.AI, you get seamless access to all popular large language models - without compromising your identity or data privacy.

Try Covertly.AI today for free at www.covertly.ai, or contact us to learn more about custom privacy and security solutions for your business.  

Works Cited

“AWS doubles down on custom LLMs with features meant to simplify model creation.” Yahoo Tech, 3 Dec. 2025, https://tech.yahoo.com/ai/gemini/articles/aws-doubles-down-custom-llms-163000225.html. 

“AWS doubles down on custom LLMs with features meant to simplify model creation.” Benzatine, 3 Dec. 2025, https://benzatine.com/news-room/aws-unveils-enhanced-tools-for-custom-ai-model-development.

Szkutak, Rebecca. “AWS doubles down on custom LLMs with features meant to simplify model creation.” TechCrunch, 3 Dec. 2025, https://techcrunch.com/2025/12/03/aws-doubles-down-on-custom-llms-with-features-meant-to-simplify-model-creation/. 

Keel, Fletcher. “What is AWS? Here’s Why So Many Sites, Apps Were Impacted During Monday’s Outage.” WLWT, 20 Oct. 2025, https://www.wlwt.com/article/what-is-aws-outage-amazon-web-services-explainer/69096520. 

Amazon Web Services. “Amazon SageMaker AI.” Amazon Web Services, https://aws.amazon.com/sagemaker/ai/.