Claude Sonnet 5.5 Guide

Claude Sonnet 5.5 is Anthropic’s latest Sonnet model, released on September 28, 2026. It is the second model in the Claude 5.5 family after Claude Opus 5.5.

It is designed to deliver a strong balance of intelligence, speed, and cost for everyday professional work.

Compared with Sonnet 5, Anthropic says Sonnet 5.5 generates outputs more than 30% faster and can cost up to 30% less per completed task.

It is particularly strong at coding, knowledge work, document creation, spreadsheets, presentations, visual understanding, computer use, and iterative workflows.

• Claude.ai
• Claude mobile apps
• Claude Code
• Claude API
• Amazon Web Services
• Google Cloud
• Microsoft Azure

API model ID

claude-sonnet-5-5

Per 1 million tokens:

Input tokens

$2

Output tokens

$10

Cache reads

$0.20

Cache writes

$2.50

Prompt caching can substantially reduce repeated-context costs. Batch processing can provide up to 50% savings for eligible asynchronous workloads.

Use Sonnet 5.5 to:

• Understand large codebases
• Build new features
• Fix bugs
• Refactor code
• Write tests
• Review code
• Debug errors
• Create applications
• Build websites
• Work across multiple files
• Handle command-line tasks
• Iterate rapidly on software

Use it to:

• Analyze reports
• Summarize research
• Compare documents
• Extract insights
• Synthesize large amounts of information
• Prepare business analyses
• Review financial information
• Create executive summaries
• Analyze charts
• Turn raw information into deliverables

Use it for:

• Reports
• Proposals
• Strategy documents
• SOPs
• Research briefs
• Meeting documents
• Marketing plans
• Business plans
• Documentation
• Executive summaries

Sonnet 5.5 is designed to produce clearer writing that requires less cleanup than previous Sonnet generations.

Use it to:

• Create slide structures
• Follow existing templates
• Write slide copy
• Transform reports into decks
• Create executive presentations
• Improve presentation narratives
• Polish slide content
• Analyze source material before building slides

Use it to:

• Analyze spreadsheet data
• Build spreadsheet structures
• Create formulas
• Clean data
• Find patterns
• Explain calculations
• Perform business analysis
• Build financial models
• Generate spreadsheet-based reports

Sonnet 5.5 can reason about visual inputs such as:

• Screenshots
• Charts
• Graphs
• Interfaces
• Documents
• Images
• Dashboards
• Visual layouts

Anthropic reports major improvements in chart recognition and long-horizon visual tasks.

Sonnet 5.5 can support agentic workflows involving computer interfaces. Potential workflows include:

• Browser tasks
• Research
• Data entry
• Application workflows
• Business operations
• Testing
• Information collection
• Multi-step digital processes

Computer-use systems should still include appropriate permissions, verification, and human oversight for consequential actions.

Sonnet 5.5 has improved design capabilities. Use it for:

• Landing pages
• Web interfaces
• Dashboards
• Frontend components
• UI improvements
• Visual hierarchy
• Layouts
• Responsive interfaces
• Design iterations
• Presentation design

Give Claude screenshots, references, design requirements, and constraints for stronger results.

Sonnet 5.5 can work through multi-step tasks involving:

• Planning
• Reasoning
• Tool usage
• Code execution
• File manipulation
• Research
• Iteration
• Verification
• Computer interaction

For production agents, define permissions, stopping conditions, validation rules, and escalation points.

Higher effort can improve difficult-task performance, but it can also increase latency and cost. More effort is not automatically better for every task.

Anthropic reports the following Sonnet 5.5 results:

Terminal-Bench 4.0

70.6% at Max effort

FrontierCode 1.1 Main

52.1% at Xhigh effort, 46.2% at Max effort

CursorBench 4.0

55.5%

GDPval-AA v2.1

1844

AA-Briefcase v1.1

1811

Humanity’s Last Exam with tools

64.5%

OSWorld 2.1

80.1% partial

Chartography

61.6% without tools

Benchmark results depend on evaluation methodology, effort settings, tools, scaffolding, and other conditions, so they should not be treated as universal real-world performance scores.

A strong prompt can contain these components.

Role: You are an expert [ROLE]. Objective: Your goal is to [OBJECTIVE]. Context: [CONTEXT] Inputs: [INPUTS] Requirements: [REQUIREMENTS] Constraints: [CONSTRAINTS] Process: [IMPORTANT STEPS OR CHECKS] Output: Produce [EXACT DELIVERABLE]. Quality criteria: The result must be accurate, practical, specific, logically consistent, and directly usable. Before finalizing: Check your work against every requirement and correct important omissions or inconsistencies.

• Be specific.
• Define the desired outcome instead of giving vague instructions.
• Provide relevant context.
• Specify the intended audience.
• Define the output format.
• Provide examples when style or structure matters.
• Separate instructions from source material clearly.
• State important constraints explicitly.
• Break very complex projects into logical stages when useful.
• Give Claude access to the information required to solve the task.
• Ask it to distinguish facts, assumptions, and uncertainties.
• Request verification when accuracy matters.
• Use structured sections for long prompts.
• Avoid contradictory instructions.
• Do not increase reasoning effort unless the task benefits from it.
• Review important outputs before acting on them.

• Build and debug software
• Analyze business documents
• Conduct research synthesis
• Create presentations
• Analyze spreadsheets
• Create marketing campaigns
• Build websites and interfaces
• Automate repetitive workflows
• Analyze screenshots, charts, and visual information
• Power AI agents and internal business tools

• Audience research
• Customer research
• Competitor analysis
• Content strategy
• SEO planning
• Ad concepts
• Landing-page copy
• Email campaigns
• Social content
• Positioning
• Campaign analysis
• Customer feedback analysis
• Content repurposing
• Marketing reports

• Market research
• Business models
• Product strategy
• Customer discovery analysis
• Competitive research
• Pitch preparation
• Financial analysis
• SOP creation
• Hiring materials
• Product requirements
• Growth experiments
• Decision-support documents

• Feature development
• Debugging
• Codebase exploration
• Code review
• Testing
• Refactoring
• Documentation
• Architecture analysis
• Frontend development
• Terminal tasks
• Repository-wide changes
• Agentic coding workflows

• Content research
• Scripts
• Newsletters
• Social posts
• Content repurposing
• Idea generation
• Research synthesis
• Content calendars
• Audience analysis
• Digital products
• Course material
• Visual-content planning

• Reports
• Documents
• Presentations
• Spreadsheets
• Research
• Support workflows
• Operations
• Data analysis
• Internal knowledge
• Process documentation
• Workflow automation

Choose Sonnet 5.5 when you want:

• Strong capability
• Fast responses
• Lower API costs
• High-volume professional work
• Coding
• Document creation
• Design work
• Business analysis
• Repeated AI workflows
• Agentic tasks with controlled complexity

Consider Opus 5.5 when the task involves:

• Very high complexity
• Open-ended problems
• Sustained judgment
• Difficult reasoning
• Complex agentic workflows
• Cases where maximizing capability matters more than minimizing cost

• Use lower effort for simple work.
• Use prompt caching for repeated context.
• Use batch processing where appropriate.
• Avoid repeatedly sending irrelevant context.
• Keep reusable instructions stable.
• Route simple tasks to lower-cost workflows.
• Reserve higher effort for difficult tasks.
• Measure cost per completed task, not just cost per token.
• Track input tokens, output tokens, tool calls, retries, latency, and success rate.

• Give Claude the actual source material.
• Explain the business objective.
• Specify what success means.
• Provide examples of strong outputs.
• Define constraints early.
• Ask for evidence when making factual claims.
• Require uncertainty to be stated instead of guessed away.
• Use iteration for creative work.
• Use explicit acceptance criteria for coding.
• Give agents clear tool permissions.
• Require verification before consequential actions.
• Keep humans involved in high-impact decisions.

Claude Sonnet 5.5 can still:

• Make factual mistakes
• Misinterpret ambiguous instructions
• Generate incorrect code
• Make calculation errors
• Misread incomplete context
• Produce plausible but unsupported claims
• Fail during long multi-step workflows
• Use tools incorrectly
• Overlook edge cases

AI-generated output should therefore be verified when used for financial, legal, medical, security, business-critical, or other consequential decisions.

• Sonnet 5.5 introduces stronger safeguards for certain high-risk cybersecurity requests.
• Anthropic says routine software development and ordinary bug fixing remain supported.
• It uses biology safeguards targeting certain high-risk requests.
• It also includes protections intended to reduce large-scale model distillation and reasoning extraction.

• Define the outcome.
• Give Claude the necessary context.
• Attach relevant files, data, images, or code.
• State constraints.
• Define the required deliverable.
• Choose an appropriate effort level.
• Let Claude complete the task.
• Review factual claims and important decisions.
• Give targeted feedback.
• Iterate until the output meets your acceptance criteria.

Need fast everyday professional work?

Use Sonnet 5.5.

Need coding, docs, decks, or spreadsheets?

Use Sonnet 5.5.

Need high-volume workflows at low cost?

Use Sonnet 5.5 with lower or medium effort.

Need routine summaries or classification?

Use lower effort.

Need content creation or standard coding?

Use medium effort.

Need complex debugging or architecture?

Use higher effort.

Need repeated context at low cost?

Use prompt caching.

Need async work at up to 50% savings?

Use batch processing.

Need very complex, open-ended work?

Consider Opus 5.5.