Anthropic Launches Claude Opus 4.7 on Amazon Bedrock: Next-Gen AI for Enterprise Coding and Agents

Breaking: Anthropic Releases Claude Opus 4.7 on Amazon Bedrock

Anthropic today announced the availability of Claude Opus 4.7 on AWS Amazon Bedrock, marking a significant upgrade in AI agentic coding, knowledge work, and long-running task autonomy. The model, described as Anthropic’s most intelligent Opus version, is powered by Bedrock’s new inference engine designed for enterprise-grade production workloads.

“Claude Opus 4.7 represents a leap forward in AI’s ability to handle complex, multi-step reasoning and code generation at scale,” said Anthropic Vice President of Product, Michelle Lee. “Combined with Amazon Bedrock’s zero-operator access, enterprises can deploy this model with confidence in sensitive environments.”

Key Improvements and Benchmarks

The model delivers strong performance across several domains, particularly in agentic coding. On SWE-bench Pro, Opus 4.7 scored 64.3%, on SWE-bench Verified 87.6%, and on Terminal-Bench 2.0 69.4%—extending the lead of its predecessor, Opus 4.6.

For knowledge work, the model achieves 64.4% on Finance Agent v1.1, excelling in document creation, financial analysis, and multi-step research. It reasons through underspecified requests, makes sensible assumptions, and self-verifies outputs to improve first-pass quality.

Long-Running Tasks and Vision

Claude Opus 4.7 maintains performance over its full 1 million token context window for long-running agents, staying on track through ambiguity. It also adds high-resolution image support for charts, dense documents, and screen UIs.

“The vision capabilities are a game-changer for analyzing complex documents and screen captures,” commented Dr. Sarah Chen, AI Researcher at Anthropic. “Accuracy on fine-detail tasks has improved substantially.”

Background

Amazon Bedrock is a managed service that provides foundation models from multiple AI companies through a single API. Anthropic’s Claude Opus series is designed for advanced reasoning and code generation, competing with models like OpenAI’s GPT-4 and Google’s Gemini. This release builds on Bedrock’s new inference engine, which uses dynamic scheduling and scaling to improve availability for steady-state and rapidly scaling workloads.

The new engine also enforces zero-operator access, meaning customer prompts and responses are never visible to Anthropic or AWS operators—a critical feature for regulated industries such as finance and healthcare.

What This Means

For developers and enterprises, Claude Opus 4.7 on Bedrock enables more reliable, autonomous coding agents that can work over longer horizons without losing context. The combination of improved reasoning, vision, and secure infrastructure reduces the need for frequent human intervention in production workflows.

“This is the first model that we feel confident running 100,000 requests per second across multiple regions,” said Tom Richards, CTO of a fintech early adopter. “The scheduling logic in Bedrock’s inference engine made that possible without privacy trade-offs.”

The model is available now in the Amazon Bedrock console via the Playground or programmatically through the Anthropic Messages API and Bedrock’s Converse API. Anthropic recommends consulting its prompting guide to optimize results, as the model may require tweaks compared to Opus 4.6.

• Availability: Amazon Bedrock console, Playground, and programmatic APIs.
• Performance: 64.3% SWE-bench Pro, 87.6% SWE-bench Verified, 69.4% Terminal-Bench 2.0, 64.4% Finance Agent v1.1.
• Security: Zero-operator access, private prompt/response handling.
• Context window: 1 million tokens.

“We’re just scratching the surface of what agentic AI can do in enterprise settings,” concluded Anthropic’s Lee. “Claude Opus 4.7 on Bedrock is a major step toward fully autonomous, secure, and scalable AI workflows.”