Anthropic’s fast, agentic model for coding, tool use, computer workflows, and everyday professional work.
Agentic coding
Completes complex implementations, debugging, refactoring, and multi-step engineering tasks
Autonomous execution
Plans work, uses tools, and carries tasks through to tested results
1M token context
Processes large codebases, documents, and extended project histories
128K token output
Produces detailed implementations, reports, and long-form deliverables
Adaptive thinking
Adjusts reasoning effort based on task complexity
Computer use
Operates browsers, terminals, and graphical interfaces
Advanced tool use
Coordinates external tools, APIs, and development environments
Multimodal understanding
Processes text, images, screenshots, charts, and PDF documents
Self-verification
Tests its own changes and checks results before completing tasks
Cost-efficient performance
Approaches Opus-class performance on some tasks at a lower price
Real-time security safeguards
Applies additional protections to high-risk cybersecurity requests
- Build and refactor applications across large codebases
- Run autonomous and semi-autonomous coding agents
- Debug complex software issues and generate tested fixes
- Perform repository-wide migrations and architectural changes
- Create production-ready interfaces from requirements and screenshots
- Automate browser, terminal, and computer-based workflows
- Analyze large documents, PDFs, charts, and technical specifications
- Conduct research and everyday professional knowledge work
- Build AI agents that coordinate tools and external services
- Automate business operations and multi-step enterprise workflows
- Generate structured reports, technical plans, and implementation strategies
- Replace Claude Sonnet 4.6 in existing API applications
Available through the Claude API with usage-based pricing.
Introductory API pricing through August 31, 2026:
Input: $2 per million tokens
5-minute prompt cache write: $2.50 per million tokens
1-hour prompt cache write: $4 per million tokens
Prompt cache hits and refreshes: $0.20 per million tokens
Output: $10 per million tokens
Standard API pricing starting September 1, 2026:
Input: $3 per million tokens
5-minute prompt cache write: $3.75 per million tokens
1-hour prompt cache write: $6 per million tokens
Prompt cache hits and refreshes: $0.30 per million tokens
Output: $15 per million tokens
Batch API pricing through August 31, 2026:
Input: $1 per million tokens
Output: $5 per million tokens
The full 1M-token context window is available at standard token rates without a separate long-context premium.
Claude Sonnet 5 is available across Claude Free, Pro, Max, Team, and Enterprise plans. Usage limits vary by plan.
Commercial usage is subject to Anthropic’s applicable terms, service policies, and usage policies.
Integration highlights
- Claude API
- Claude.ai
- Claude Code
- Claude Platform
- Claude Cowork
- Model Context Protocol (MCP)
- Amazon Bedrock
- Claude Platform on AWS
- Google Cloud Vertex AI
- Microsoft Foundry
- VS Code
- JetBrains
- Slack
- Microsoft Excel
- Microsoft PowerPoint
- Microsoft Word