AWS Weekly Roundup: Graviton5, Gemma 4 on Bedrock, and OpenSearch AI Updates

Amazon Web Services has released its weekly roundup, detailing several key updates across its compute, machine learning, and observability services. A major highlight is the general availability of the new Graviton5-powered Amazon EC2 M9g and M9gd instances. These instances, built on the Graviton5 processors and the sixth-generation AWS Nitro System, deliver up to 25% better compute performance compared to their Graviton4 predecessors. Specific improvements include up to 35% faster performance for web applications and machine learning inference, and up to 30% for databases. Graviton5 is notable for being the first AWS processor to support PCIe Gen6 and DDR5-8800 memory, and it includes a 5x larger L3 cache. This release also introduces the Nitro Isolation Engine, enhancing virtual machine isolation through formal verification. In the realm of artificial intelligence, Google DeepMind's Gemma 4 family of models is now available on Amazon Bedrock. This includes three variants: Gemma 4 31B, Gemma 4 26B-A4B (a mixture-of-experts architecture), and Gemma 4 E2B (the smallest variant for low-latency use cases). All Gemma 4 models support native function calling, structured output, reasoning, response streaming, and multimodal input across text, image, video, and audio, supporting over 35 languages. This integration significantly expands the generative AI options available to AWS customers. Furthermore, Amazon OpenSearch Service has launched MCP Apps, which introduce agentic observability capabilities. These applications enable AI agents within compatible IDEs like Claude Desktop and VS Code to investigate incidents. Agents can leverage logs, traces, metrics, and alerts stored in OpenSearch domains, collections, and Amazon Managed Service for Prometheus. Each MCP App tool call provides both a text summary for the agent's reasoning and an interactive visualization within the same conversation thread, streamlining incident analysis. Available tools cover various aspects of log, metrics, and trace investigation, service performance, and dynamic visualizations. The roundup also included an important note regarding Anthropic's Claude Fable 5 and Mythos 5 models. On June 12, Anthropic requested AWS to revoke access to these models for all users to comply with a US Government export control directive, though other models like Opus 4.8 remain unaffected. AWS stated it would provide further updates as they become available. This week's announcements underscore AWS's continuous efforts to enhance its core infrastructure and expand its AI/ML offerings, providing developers and enterprises with more powerful tools for their cloud-native applications and AI-driven initiatives.