Amazon Nova Forge: "Open training” paradigm that empowers everyone to build their own frontier AI
Amazon just opened the training process. Now customers can build frontier AI with their own data at every stage.
Why it matters
Amazon Nova Forge democratizes frontier AI development by allowing enterprises to inject proprietary data into model training at multiple checkpoints, solving the domain adaptation problem without catastrophic forgetting—a significant shift in how companies can customize large language models.
The key facts
5 to knowAmazon Nova Forge enables customers to mix proprietary data with Amazon Nova training data at each major stage of model development
Service addresses catastrophic forgetting problem in model fine-tuning
Positions Amazon as enabling 'open training' paradigm for frontier AI customization
Allows deep domain understanding integration into foundation models
Launched December 8, 2025
Go to the source
Amazon Scienceamazon.science
Publisher excerpt: New service lets customers mix their own data with the data used to train Amazon Nova at each major stage of model development, enabling deep domain understanding while preventing "catastrophic forgetting".