Claude Opus 4
Hybrid reasoning model that pushes the frontier for coding and AI agents, featuring a 200K context window
Announcements
- New
Claude Opus 4
May 22, 2025
Claude Opus 4 is our most intelligent model to date, pushing the frontier in coding, agentic search, and creative writing. We’ve also made it possible to run Claude Code in the background, enabling developers to assign long-running coding tasks for Opus to handle independently.
Read more
Availability and pricing
For business users and consumers who want to collaborate with our most powerful model on complex tasks, Claude Opus 4 is available on Claude for Pro, Max, Team, and Enterprise users.
For developers interested in building AI solutions that demand frontier intelligence, Claude Opus 4 is available on the Anthropic API, Amazon Bedrock, and Google Cloud's Vertex AI.
Pricing for Claude Opus 4 starts at $15 per million input tokens and $75 per million output tokens, with up to 90% cost savings with prompt caching and 50% cost savings with batch processing. To learn more, check out our pricing page.
Use cases
Claude Opus 4 is our most intelligent model to date, pushing the frontier in coding, agentic search, and creative writing. With advanced reasoning and powerful collaboration capabilities, Claude Opus 4 transforms how teams approach complex engineering and business challenges.
Claude Opus 4 offers hybrid reasoning that allows for instant responses or extended, step-by-step thinking visible through user-friendly summaries. API users have fine-grained control over thinking budgets for optimal cost and performance. Popular use cases include:
AI agents
Claude Opus 4 delivers state-of-the-art performance on complex agent applications, with strong results on TAU-bench and exceptional accuracy for long-horizon tasks. From autonomously managing multi-channel marketing campaigns to orchestrating cross-functional enterprise workflows, Claude Opus 4 powers sophisticated agent architectures that expand AI capabilities for your products and services.
Advanced coding
Claude Opus 4 is a leader on SWE-bench, completing days-long engineering tasks in coherent, context-aware solutions across thousands of steps. With improved code taste and 32K output token support, it adapts to specific coding styles while delivering exceptional quality for extensive generation and refactoring projects.
Agentic search and research
Claude Opus 4 can effectively search through external and internal data sources to synthesize comprehensive insights across complex information landscapes. With powerful performance on agentic search, it can conduct hours of independent research—simultaneously analyzing everything from patent databases to academic papers and market reports, delivering strategic insights for decision-making.
Content creation
With rich, deep character and excellent writing abilities, Claude Opus 4 produces human-quality content and more natural, prose-focused outputs. It outperforms previous Claude models on creative writing, in addition to exceptional visual taste.
Benchmarks
Claude Opus 4 delivers frontier intelligence across coding, agentic search, and AI agent capabilities.
Claude Opus 4 achieves industry-leading results on SWE-bench for coding. It demonstrates strong performance on MMLU, GPQA, and Aider Polyglot. With exceptional writing abilities, Claude Opus 4 surpasses previous Claude models and new standards for natural, human-like prose.
Trust & Safety
We've conducted extensive testing and evaluation of Claude Opus 4, working with external experts to ensure it meets our standards for safety, security and reliability. In the model card for this release, we discuss new safety results in several categories.
What customers are saying
Claude Opus 4 and Sonnet 4 are state of the art coding models. They're a leap forward in complex codebase understanding, and we expect developers will experience across the board capability improvements.
Claude Opus 4 offers truly advanced reasoning for coding. When our team deployed it on a complex open source project, it coded autonomously for nearly seven hours—a huge leap in AI capabilities that left the team amazed.
Claude Opus 4 and Sonnet 4 push the envelope once again on coding capabilities, handling complex multi-file changes without touching code you didn't ask to modify. The improved precision in instruction following plus thinking mode gains show real potential to fundamentally change how Replit Agent works.
Anthropic has again set the gold standard for code-generation models with Claude Opus 4 and Sonnet 4. With a modest shift in prompting, we've seen the new models deliver cleaner, more precise high-quality output.
Claude Opus 4 excels at solving complex challenges that other models can't. It successfully handles critical actions that previous models have missed—that kind of reliability is transformative for our development workflows.
We are excited for Claude Opus 4's frontier intelligence capabilities, especially around agentic reasoning and tool use. Our customers will be able to build even better agents on Databricks that use it to reason over their enterprise data.
Enterprises rely on Snowflake to extract insights from data using AI. With Claude Opus 4—offering custom tool instructions and advanced multi-hop reasoning—we’re expanding the capabilities of data agents in Snowflake Cortex AI. This unlocks greater value from unstructured data and empowers customers to build intelligent, agentic applications for real-world business challenges.
Claude Opus 4 is the first model that boosts code quality during editing and debugging in its agent, codename goose, without sacrificing performance or reliability.
Claude Opus 4 shone on one of the most complex litigation tasks I can think of: engaging a full litigation record and populating a summary judgment with granular citations mapped to specific elements of cause of action.
When it comes to accuracy in agentic systems and enterprise datasets—particularly those requiring tool use and multi-turn interaction—Claude Opus 4 is advancing the state-of-the-art. Our researchers benchmarked it for a real world insurance underwriting use case using Snorkel’s fine-grained evaluation platform and saw it significantly outperform other reasoning models on critical subsets of data like specific lines of business.
Claude Opus 4 excels for text-to-SQL use cases—beating internal benchmarks as the best model we've tried. We're switching all of our agent workloads to it to leverage this superior reliability and intelligence.
Against our in-house benchmark, Claude Opus 4 was a dramatic improvement in code generation abilities compared to Anthropic's previous models. We're incredibly excited to see such progress, and for the product opportunities this unlocks.
Claude Opus 4 brings advanced reasoning and agentic capabilities that significantly enhance our flagship AI agent, Archie. It delivers a leap in performance for financial analysis on complex Excel files, decks, and charts.
Claude Opus 4 strikes a balance of intelligence and taste which we've found unmatched in other models. Its capabilities combined with its human-like character make it a genuine pleasure to work with!
Frequently asked questions
We offer a family of Claude models across the spectrum of speed, price, and performance. Claude Opus 4 is our most intelligent model to date. We recommend Claude Opus 4 for your most demanding use cases where you need frontier intelligence, such as complex AI agents, advanced coding projects, deep research tasks, and long-horizon autonomous work where accuracy and capability matter more than speed or cost.
Pricing depends on how you want to use Claude Opus 4. To learn more, check out our pricing page.
Claude Opus 4 is both a standard model and a hybrid reasoning model in one: you can pick when you want the model to answer normally and when you want it to utilize extended thinking.
Extended thinking mode is best for use cases where performance and accuracy matter more than latency. It significantly improves response quality for complex reasoning tasks, extended agentic work, multi-step coding projects, and deep research—and the thinking summaries help you understand key aspects of the model's reasoning process.