# Anthropic Introduces Claude 4 Sonnet and Claude 4 Opus: A Major Progression in AI Functionality
Following significant tech gatherings like Microsoft Build and Google I/O, Anthropic has announced the debut of Claude 4 Sonnet and Claude 4 Opus. These models are now accessible on Claude’s website and via the API, aiming to improve AI capabilities across various sectors, especially in coding and reasoning.
## Superior Performance in Coding and Gaming
According to Anthropic, Claude Sonnet 4, which sits as a mid-tier model between Raiku and Opus, exhibits outstanding advancements in coding, reasoning, and following instructions when compared to its predecessor, Claude Sonnet 3.7. This upgrade is vital as developers increasingly seek AI models capable of tackling intricate coding challenges.
Conversely, Claude Opus 4 is said to equal or even exceed the performance of top models like OpenAI’s o3, GPT-4.1, and Google’s Gemini 2.5 Pro across numerous benchmarks. These benchmarks cover multilingual question-and-answer tasks, agentic tool application, and advanced reasoning. This is particularly notable since Claude had previously been a go-to choice among developers for coding but had recently lagged due to updates from rivals.
In an intriguing contrast, while Google’s Gemini 2.5 Pro completed a full session of Pokémon Blue, Claude Opus 4 showcased its skills by engaging with Pokémon for an impressive 24 hours, a substantial improvement from the 45 minutes accomplished by its older version.
## New Features and Functionalities
Alongside the introduction of these new models, Anthropic rolled out several additional features that enhance Claude’s functionality:
– **Extended Reasoning with Tool Use (Beta)**: Both models can leverage tools like web searches during elaborate reasoning sessions. This enables Claude to switch between reasoning and tool utilization, resulting in more accurate and detailed responses.
– **Enhanced Model Capabilities**: The new models are equipped to use tools in tandem, follow instructions more accurately and, when given access to local files, exhibit enhanced memory functions. This allows them to extract and retain essential information over time, improving continuity in interactions.
– **Claude Code Availability**: After a fruitful research preview, Claude Code is now widely available. This feature facilitates better collaboration between developers and Claude, supporting background tasks through GitHub Actions and integrating smoothly with development environments like VS Code and JetBrains. This integration fosters real-time editing and pair programming, optimizing the development workflow.
– **Updated API Capabilities**: Anthropic has introduced four new features on its API, allowing developers to build more advanced AI agents. These include a code execution tool, an MCP connector, a Files API, and the capability to cache prompts for up to one hour.
These developments are especially advantageous for developers, as they can now reference Claude directly in GitHub pull requests, promoting a more effective and collaborative coding experience.
## Pricing and Accessibility
Both Claude 4 Sonnet and Claude 4 Opus can be accessed through the Anthropic API and partner platforms like Amazon Bedrock and Google Cloud’s Vertex AI. The pricing is structured as follows:
– **Claude Opus 4**: $15/$75 per million tokens (input/output)
– **Claude Sonnet 4**: $3/$15 per million tokens (input/output)
## Summary
The introduction of Claude 4 Sonnet and Claude 4 Opus represents a significant advancement for Anthropic in the competitive arena of AI models. With enhanced coding capabilities, improved reasoning, and new features tailored for developers, these models are set to become invaluable resources across various applications. As AI technology continues to advance, the influence of these innovations will likely be significant across multiple industries, making this an exciting time for both developers and users.
Are you currently utilizing Claude or similar large language models in your work? Share your experiences in the comments!
