Claude Opus 4 is Anthropic’s most advanced AI model to date. It’s capable of handling complex tasks that require deep thinking and extended focus. For instance, in tests, Opus 4 successfully played the game Pokémon Red for 24 hours straight, showcasing its ability to manage long-term tasks without losing track .WIRED
In terms of coding, Opus 4 excels. It achieved top scores on industry benchmarks, such as 72.5% on SWE-bench and 43.2% on Terminal-bench, outperforming competitors like OpenAI’s GPT-4.1 and Google’s Gemini 2.5 Pro .
A unique feature of Opus 4 is its “hybrid reasoning” capability. This allows the model to switch between quick responses and deeper, more thoughtful analysis, depending on the task at hand.
Claude Sonnet 4: Versatile and Accessible AI
Claude Sonnet 4 is designed for a wide range of everyday tasks. It’s an upgrade from the previous Sonnet 3.7 model, offering improved performance in areas like coding, reasoning, and following instructions .
One of the key advantages of Sonnet 4 is its accessibility. While Opus 4 is available to paying users, Sonnet 4 can be accessed by both free and paid users, making advanced AI tools more widely available .WIRED
Developers have praised Sonnet 4 for its capabilities. For example, GitHub plans to integrate it into their Copilot coding assistant, citing its strong performance in agentic scenarios .
New Features and Tools for Developers
Anthropic has also introduced several new tools to enhance the capabilities of the Claude 4 models:
- Code Execution Tool: Allows the AI to run code, enabling interactive problem-solving.
- MCP Connector: Standardizes context exchange between AI assistants and software environments.
- Files API: Enables the AI to work directly with files, improving its ability to handle real-world tasks.
- Prompt Caching: Allows developers to cache prompts for up to an hour, improving efficiency for repeated queries.
These tools are accessible via the Anthropic API, Amazon Bedrock, and Google Cloud’s Vertex AI, ensuring broad availability for developers and enterprises .
Pricing and Availability
Anthropic has maintained consistent pricing for its models:
- Claude Opus 4: $15 per million input tokens and $75 per million output tokens.
- Claude Sonnet 4: $3 per million input tokens and $15 per million output tokens .
Notably, Claude Sonnet 4, including its extended thinking capabilities, is available to free-tier users, making advanced AI more accessible to a wider audience .
Commitment to Safety and Responsible AI
Given the advanced capabilities of Claude Opus 4, Anthropic has implemented heightened safety measures, categorizing it under AI Safety Level 3 (ASL-3). This includes enhanced cybersecurity protocols and anti-jailbreak measures, underscoring Anthropic’s commitment to responsible AI deployment and risk mitigation .
With the launch of Claude Opus 4 and Claude Sonnet 4, Anthropic is setting new standards in AI performance, offering powerful tools for developers, businesses, and researchers alike. These models not only advance the state of AI technology but also emphasize the importance of safety and accessibility in AI development.
