On the heels of Microsoft Build and Google I/O, Anthropic has just announced Claude 4 Sonnet and Claude 4 Opus, which are immediately available on Claude’s website, as well as in the API. Here’s what’s new.
Better at coding and at … Pokémon
According to Anthropic, Claude Sonnet 4 (its mid-tier model, between Raiku and Opus) significantly improves at coding, reasoning, and instruction following compared to its predecessor, Claude Sonnet 3.7.
As for Claude Opus 4, Anthropic says it matches or outperforms OpenAI’s o3, GPT-4.1, and Gemini 2.5 Pro in benchmarks for multilingual Q&A, agentic tool use, agentic terminal coding, agentic coding, and graduate-level reasoning:
This is especially significant because, while Claude spent most of last year at the top of developers’ preferred models for coding tasks, it has fallen behind in recent weeks after multiple model updates by OpenAI and Google.
And speaking of Google, its Gemini 2.5 Pro model made the rounds recently after it completed a Pokémon Blue playthrough. Anthropic was happy to report that while it hasn’t yet achieved the same feat, Claude Opus 4 was able to agentically play Pokémon for 24 hours, versus 45 minutes from the previous version.
New capabilities and Claude Code
Alongside the models, Anthropic also announced:
• Extended thinking with tool use (beta): Both models can use tools—like web search—during extended thinking, allowing Claude to alternate between reasoning and tool use to improve responses.
• New model capabilities: Both models can use tools in parallel, follow instructions more precisely, and—when given access to local files by developers—demonstrate significantly improved memory capabilities, extracting and saving key facts to maintain continuity and build tacit knowledge over time.
• Claude Code is now generally available: After receiving extensive positive feedback during our research preview, we’re expanding how developers can collaborate with Claude. Claude Code now supports background tasks via GitHub Actions and native integrations with VS Code and JetBrains, displaying edits directly in your files for seamless pair programming.
• New API capabilities: We’re releasing four new capabilities on the Anthropic API that enable developers to build more powerful AI agents: the code execution tool, MCP connector, Files API, and the ability to cache prompts for up to one hour.
The Claude Code news is particularly interesting for developers, since @ mentioning Claude and letting it run directly from a GitHub PR has the potential to streamline the development process.
Anthropic says both models are available on the Anthropic API and partners like Amazon Bedrock and Google Cloud’s Vertex AI. Opus 4 costs $15/$75 per million tokens (input/output), and Sonnet 4 costs $3/$15 per million tokens (input/output).
Do you use Claude or other LLMs at work? Let us know in the comments.
FTC: We use income earning auto affiliate links. More.
