Amazon aggiorna AWS Bedrock con nuova capacità di importare modelli AI personalizzati.

Amazon Web Services Introduces New Capabilities for Bedrock AI Service

Amazon Web Services is expanding the capabilities of its Bedrock artificial intelligence service, allowing customers to import their own custom AI models alongside those from major providers.
This move is part of a strategic effort to attract AI application developers by providing access to a diverse range of AI models through the Bedrock managed service.

Competition in the AI Space

In the fiercely competitive AI landscape, Amazon is vying against tech giants like Microsoft and Google, as well as emerging platforms such as OpenAI and HuggingFace.
The Custom Model Import feature enables customers to seamlessly integrate their externally developed models into Bedrock, positioning them as integral components within the Bedrock workflows.

Vasi Philomin, AWS’s vice president of generative AI, emphasized the significance of this new capability, highlighting how customers can leverage their unique AI creations within the Bedrock ecosystem.

Key Updates and Enhancements

Amazon has announced several enhancements to its Bedrock service, including updates to the Titan foundation models.
The upcoming launch of the Titan Text Embeddings Model V2 promises improved accuracy and efficiency, while the Titan Image Generator now features built-in invisible watermarking for AI-generated image recognition.

Furthermore, AWS has introduced additional third-party models to Bedrock, such as Meta’s Llama 3 foundation models for generative AI applications and Cohere’s Command R and Command R+ models tailored for enterprise AI solutions.
Additionally, Anthopic’s Claude 3 Opus model has recently been released on the Bedrock platform.

Enhanced Features and Availability

Among the latest updates, Amazon has made its Model Evaluation capability widely available on Bedrock, enabling developers to compare and evaluate different models effectively.
Additionally, the introduction of the Guardrails feature enhances data security by implementing safeguards to filter out personal information, profanity, and other harmful content from AI models.

This continuous innovation and expansion of capabilities demonstrate Amazon’s commitment to empowering AI developers and enterprises with cutting-edge tools and resources within the Bedrock ecosystem.