Agentic software development
Use Sonnet 5 for repository-scale debugging, implementation, testing, and terminal or browser workflows that require several connected steps.
Independent tool overview
Claude Sonnet 5 at a glance
Claude Sonnet 5 is Anthropic’s balanced frontier model for coding, multi-step agents, research, and professional work. It combines a 1 million-token context window and up to 128,000 output tokens with permanent API pricing of $2 per million input tokens and $10 per million output tokens.
Visit the official Claude Sonnet 5 site ↗
- Best for
- Coding agents, tool-using workflows, and long-context professional work
- API price
- $2 / million input tokens and $10 / million output tokens
- Context window
- 1 million tokens
- Maximum output
- 128,000 tokens
- Model ID
- claude-sonnet-5
- Chat access
- Default on Claude Free and Pro; available on Max, Team, and Enterprise
- Last reviewed
- August 29, 2026
Overview
What Claude Sonnet 5 is
Claude Sonnet 5 is positioned between Anthropic’s faster, lower-cost Haiku models and its most capable Opus models. It is the default model for Claude Free and Pro users, and it is also available through Claude Code, the Claude API, Amazon Bedrock, Google Cloud, and Microsoft Foundry.
The model is designed for sustained agentic work: it can plan, use browsers and terminals, call tools, inspect results, and continue through multi-step coding or knowledge-work tasks. Adaptive thinking is enabled by default, letting developers trade latency and token use against reasoning depth with the effort setting.
For developers, the headline numbers are a 1 million-token context window, a 128,000-token maximum output, and the API model ID claude-sonnet-5. Those limits make Sonnet 5 useful for large repositories and document collections, but teams should still test retrieval quality and total workload cost instead of assuming every token receives equal attention.
Use cases
Who Claude Sonnet 5 is best for
The strongest fit depends on the job you need the product to complete, not the size of its feature list.
Production AI applications
Build assistants and agents that need a strong balance of capability, speed, and per-token cost without paying Opus pricing for every request.
Long-document analysis
Analyze large document sets, codebases, and multimodal inputs inside a 1 million-token context window, subject to request and image limits.
Professional knowledge work
Draft, research, synthesize, and operate across connected tools in Claude Chat, Claude Code, or custom API workflows.
Capabilities
Core Claude Sonnet 5 features
Agentic planning and tool use
Sonnet 5 can plan work, use browsers and terminals, call tools, and continue through multi-step tasks rather than stopping after a single answer.
Adaptive thinking
Reasoning is adaptive by default, with effort controls that let developers tune the balance among quality, latency, and billed output tokens.
1 million-token context
The standard context window supports up to 1 million tokens across Anthropic’s native API and supported cloud platforms.
Up to 128K output
A maximum output of 128,000 tokens supports long reports, code generation, and complex agent traces when the use case justifies it.
Browser and computer use
The Claude API supports browser use and Anthropic’s computer-use toolset for workflows that must interact with websites or desktop-style interfaces.
Files, PDFs, and vision
Sonnet 5 works with text, images, PDFs, and files, making it suitable for multimodal document and interface analysis.
Production platform support
Developers can use Sonnet 5 through the Claude API, Amazon Bedrock, Google Cloud, and Microsoft Foundry.
Caching and batch options
Prompt caching and batch processing can reduce repeat-input and asynchronous workload costs, depending on the request pattern.
Process
How the Claude Sonnet 5 workflow works
Step 1
Choose Chat or API access
Use a Claude subscription for interactive work, Claude Code for coding sessions, or the metered API for software and automated agents.
Step 2
Benchmark a representative task set
Test Sonnet 5 on real prompts, tools, repositories, and failure cases rather than relying only on model benchmarks.
Step 3
Update migration settings
Switch to claude-sonnet-5, replace manual thinking budgets with adaptive thinking and effort, and remove unsupported custom sampling parameters.
Step 4
Recalculate token economics
Measure requests with the updated tokenizer because the same text can use roughly 1.0 to 1.35 times as many tokens as Sonnet 4.6.
Step 5
Add operational guardrails
Set spend limits, tool permissions, approval points, timeouts, and validation checks before allowing autonomous actions.
Step 6
Monitor quality and refusals
Track task success, latency, token use, tool errors, and responses whose stop reason is refusal, including HTTP 200 responses.
Cost
Claude Sonnet 5 pricing and free plan
Claude subscriptions and Claude API usage are billed separately. Sonnet 5 API pricing is permanently $2 per million input tokens and $10 per million output tokens; subscription limits and taxes vary.
Claude Free
$0
Limited Claude Chat access with Sonnet 5 as the default model.
- Web, desktop, iOS, and Android access
- Usage limits apply
- Includes web search, files, code execution, and extended thinking
Claude Pro
$20/month or $200/year
Individual plan with more usage and access to Claude Code, Cowork, projects, Research, and additional models.
- Annual price is $200 billed up front
- Equivalent advertised annual rate is about $17/month
- Usage limits apply
Claude Max
$100/month or $200/month
Individual plans offering approximately 5x or 20x Pro usage per session.
- Higher output limits
- Priority access at busy times
- Billed monthly
Claude Team
From $20/seat/month annually
Team workspace for organizations with 2 to 150 seats.
- Standard seat: $20 annually or $25 monthly
- Premium seat: $100 annually or $125 monthly
- Central administration and SSO included
Claude Enterprise
$20/seat/month plus API-rate usage
Enterprise deployment with advanced administration, security, and metered model usage.
- Billed annually
- Usage cost depends on model and task
- Custom controls and compliance features
Sonnet 5 API
$2 input / $10 output per million tokens
Metered developer access using the claude-sonnet-5 model ID.
- Pricing is permanent as of Anthropic’s August 2026 update
- Prompt caching and batch processing may reduce qualifying costs
- Thinking tokens are billed as output tokens
Pricing checked . Check current pricing at the source ↗
Assessment
Claude Sonnet 5 strengths and limitations
Where it stands out
- Strong capability-to-cost balance for coding, agents, and professional work
- Large 1 million-token context window with up to 128,000 output tokens
- Permanent $2 input and $10 output pricing per million API tokens
- Adaptive thinking provides a practical quality, latency, and cost control
- Available in Claude Chat, Claude Code, the native API, and major cloud platforms
- Supports browser, computer, file, PDF, vision, caching, and batch workflows
What to consider
- Sonnet 5 can hallucinate, misunderstand instructions, or make incorrect tool calls, so consequential work still needs validation and human approval.
- Adaptive thinking can consume variable output tokens, making latency and cost less predictable than a fixed-response workflow.
- Its updated tokenizer can turn the same text into roughly 1.0 to 1.35 times as many tokens as Sonnet 4.6, so the lower per-token price may not translate directly into the same workload savings.
- Migration from older Sonnet models is not entirely drop-in: manual thinking budgets, nondefault sampling parameters, and assistant-message prefilling are unsupported.
- Anthropic’s Priority Tier is not available for Sonnet 5.
- Safety systems can return refusals with an HTTP 200 status and a refusal stop reason, which applications must handle explicitly.
- The reliable knowledge cutoff is January 2026, so current facts require retrieval, browsing, or another live data source.
- Claude Opus remains the stronger choice for Anthropic’s hardest tasks, while Haiku can be faster and cheaper for simpler workloads.
Compare
Claude Sonnet 5 alternatives
The right alternative depends on the specific output, workflow, controls and budget your project requires.
Consumer
Claude Opus 4.8
Choose the Opus line when maximum reasoning and agentic capability matter more than per-token cost.
Explore Claude Opus 4.8 →Business Operations
ChatGPT
Choose ChatGPT for a broad consumer and team assistant ecosystem with OpenAI models, tools, and integrations.
Explore ChatGPT →Consumer
Claude Opus 4.6
Consider an older Opus model when an existing workflow is already benchmarked and stable on that model.
Explore Claude Opus 4.6 →Questions
Claude Sonnet 5 FAQs
What is Claude Sonnet 5?
Claude Sonnet 5 is Anthropic’s balanced frontier model for coding, tool use, agents, and professional knowledge work. It sits below the Opus line in maximum capability and above Haiku in typical capability and cost.
How much does the Claude Sonnet 5 API cost?
The standard API price is $2 per million input tokens and $10 per million output tokens. Anthropic made that price permanent in August 2026.
Is Claude Sonnet 5 free?
Sonnet 5 is the default model on Claude’s Free plan, subject to usage limits. Developers using the API pay separately for token usage.
What is the context window for Claude Sonnet 5?
Sonnet 5 supports a 1 million-token context window by default and a maximum output of 128,000 tokens.
Is Claude Sonnet 5 available in Claude Code?
Yes. Anthropic makes Sonnet 5 available in Claude Code as well as Claude Chat and the Claude API.
What changes when migrating from Sonnet 4.6?
Developers should use adaptive thinking and effort instead of manual thinking budgets, remove unsupported nondefault sampling parameters, avoid assistant-message prefilling, and recalculate token use with the updated tokenizer.
Is Sonnet 5 better than Claude Opus?
Sonnet 5 offers a lower-cost balance of speed and intelligence. Anthropic positions Opus as its more capable option for the most complex agentic, coding, and enterprise tasks.
Does a Claude subscription include API usage?
No. Claude Chat subscriptions and developer API usage are separate products with separate billing.
Bottom line
Our Claude Sonnet 5 verdict
Claude Sonnet 5 is the practical default for teams that want Anthropic’s modern agentic and coding capabilities without paying Opus rates. Its 1 million-token context, broad tool support, and permanent $2/$10 API price are compelling, but migration changes, higher token counts, variable thinking costs, and the need for human review should be part of the implementation plan.
Visit Claude Sonnet 5 website ↗Similar Consumer Tools
View all Consumer AI tools →
Sakana Fugu
Sakana Fugu - Sakana’s new orchestration model nearing frontier capabilities
Higgsfield Explainer
Higgsfield Explainer - Turn any topic into a faceless explainer video
Framer 3.0
Framer 3.0 - Canvas-native agents that build and run your whole site
Hy-3
Hy-3 - Tencent's new open-source model built for cheap, reliable agent deployment
AI training for the future of work.
Get access to all our AI courses, hundreds of real-world AI use cases, live expert-led workshops, an exclusive network of AI early adopters, and more.
AI Courses
Get unlimited access to all of our current & upcoming industry-specific AI courses for the duration of your subscription.
Daily Guides
To keep up with the rapid pace of AI, our team publishes AI implementation guides daily. Our library contains 300+ practical use cases to automate real-world work.
Workshops
Join weekly, live, interactive sessions with industry leaders who are at the forefront of AI for hands-on implementation guidance and exclusive insights.
Community
Network with an exclusive community of AI-first professionals who are working smarter with AI. Learn how early adopters are using AI in their work and businesses.