Anthropic Launches Claude Opus 4 and Claude Sonnet 4

At its first developer conference in San Francisco, Anthropic announced the launch of two new language models: Claude Opus 4 and Claude Sonnet 4.

These models are part of the new Claude 4 family and bring major improvements in coding, reasoning, and the ability to work independently for long periods.

Claude Opus : The Flagship 4 Model

What is Claude Opus 4?

Claude Opus 4 is Anthropic’s most powerful and advanced AI model, designed for solving complex problems and running tasks without human help.

Top Performance:

Scored highest on SWE-bench (72.5%) and Terminal-bench (43.2%)—key benchmarks for coding and technical tasks.

Outperformed rivals like OpenAI’s GPT-4.1 and Google’s Gemini 2.5 Pro in tool use and programming.

Autonomous Abilities:

Can function without human guidance for up to 7 hours.

Played Pokémon Red independently for 24 hours, showing strong decision-making over time.

Claude Sonnet 4: Efficient and User-Friendly

What It Is: Claude Sonnet 4 is a streamlined, affordable version of the Opus model, aimed at everyday use.

Upgrades:

Replaces the earlier Claude 3.7 model.

Improved in coding skills, logical reasoning, and accuracy.

Accessibility:

Available to both free and paid users, making advanced AI more reachable for everyone.

Key Features of Claude 4 Models

Both Claude Opus 4 and Sonnet 4 introduce several new features:

• Hybrid Reasoning: The models can switch between internal problem-solving and using external tools like web search, enhancing their versatility.
• Extended Thinking Mode: A beta feature allowing users to optimize the balance between reasoning and tool use for improved performance.
• Thinking Summaries: Provides clearer insights into the models’ reasoning processes, aiding transparency.
• Improved Memory Handling: Can remember more over time when granted file access, helping with long-term tasks.
• Parallel Tool Execution: Enhances efficiency by allowing the models to use multiple tools simultaneously.
• Additionally, Anthropic has launched the Claude Code agentic command-line tool, enabling developers to delegate coding tasks directly from their terminal.

How to Access

Platforms: Available via Anthropic’s API, Amazon Bedrock, and Google Cloud’s Vertex AI.

Pricing:

Claude Opus 4 is only available to paid users ($15 per million input tokens, $75 per million output tokens).

Claude Sonnet 4 is available to all users, including those on free plans.

News Gist

Anthropic has launched Claude Opus 4 and Claude Sonnet 4, AI models with major upgrades in coding, reasoning, and autonomy.

Opus excels in benchmarks and long tasks, while Sonnet offers efficient performance for all users.

Features include hybrid reasoning, extended memory, and tool integration. Available via API and cloud platforms.