Claude Sonnet 5 launches on AWS Bedrock with Opus-level intelligence at Sonnet pricing

Anthropic has released Claude Sonnet 5 on Amazon Bedrock and Claude Platform on AWS. The model delivers what Anthropic describes as near-Opus intelligence while maintaining Sonnet-tier pricing, with promotional rates available through August 31, 2026.

Claude Sonnet 5 is positioned as an upgrade to Sonnet 4.6, targeting coding, agentic workflows, and professional work at scale. The model is available in multiple AWS regions through Amazon Bedrock, with access also provided through Claude Platform on AWS in North America, South America, Europe, and Asia Pacific.

Technical capabilities

According to Anthropic, Claude Sonnet 5 shows improved performance in multi-step reasoning tasks. The model can track task progress across stages, maintain execution plans, and resolve issues with fewer correction rounds compared to its predecessor.

For coding tasks, Anthropic claims Sonnet 5 can navigate existing codebases, implement multi-file changes, and handle extended debugging and refactoring sessions. The model is designed to support autonomous agents handling complex dependency chains and multi-step tool use.

The model includes computer use capabilities, enabling automation of browser and desktop workflows that previously required human interaction.

Deployment and access

Developers can access Claude Sonnet 5 through three methods on AWS:

1. Amazon Bedrock Console: Direct access through the Playground interface with model ID us.anthropic.claude-sonnet-5
2. Anthropic Messages API: Via the anthropic[bedrock] SDK package with bedrock-mantle endpoints
3. AWS Converse API: For unified multi-model experiences through standard AWS SDKs

All deployment methods use AWS SigV4 authentication and maintain enterprise security controls. Required IAM permissions include bedrock:InvokeModel, bedrock:InvokeModelWithResponseStream, and bedrock:CreateInference.

Pricing structure

Anthropic is offering promotional pricing through August 31, 2026, though specific per-token rates were not disclosed in the announcement. Standard Sonnet-tier pricing applies, positioning the model below Opus in cost while claiming near-Opus performance levels.

The model supports a maximum output of 4,096 tokens per request in the provided code examples, though the full context window specification was not mentioned.

Target use cases

Anthropic identifies three primary use cases:

• Financial services: Spreadsheet modeling, financial analysis, and self-auditing reporting agents
• Professional work: Document drafting, structured analysis, and report generation from unstructured sources
• Agent automation: Production-grade agents for unattended multi-step operations

The model is available immediately through Amazon Bedrock's standard deployment infrastructure, with AWS providing regional data residency and enterprise security controls.

What this means

Claude Sonnet 5's release on AWS represents Anthropic's push to balance performance with cost-effectiveness in the enterprise market. The AWS-exclusive launch, at least initially, strengthens Amazon's position in AI infrastructure by offering early access to Anthropic's latest Sonnet-tier model. The promotional pricing through mid-2026 suggests aggressive positioning against competing models from OpenAI and Google. However, the absence of disclosed context window size, specific benchmark scores, and exact pricing makes direct comparisons difficult. The emphasis on agentic workflows and computer use capabilities indicates Anthropic is targeting the growing market for autonomous AI systems in enterprise settings.