Anthropic introduces Claude Sonnet 4.5 for building complex agents; Some more upgrades released

Yesterday, Anthropic introduced the Claude Sonnet 4.5 coding model, which is claimed to be the strongest model for building complex agents.

Here are the details

Claude Sonnet 4.5

This model is state-of-the-art on the SWE-bench Verified evaluation, which measures real-world software coding capabilities. It is said to maintain focus for more than 30 hours on complex, multi-step tasks. On OSWorld, a benchmark that tests AI models on real-world computer tasks, Sonnet 4.5 now leads at 61.4%.

The model is revealed to show improved capabilities on a broad range of evaluations, including reasoning and math, while it is also said to arrive with better domain-specific knowledge and reasoning. Claude Sonnet 4.5 is claimed to be Anthropic’s most aligned frontier model yet, reducing concerning behaviors like sycophancy, deception, power-seeking, and the tendency to encourage delusional thinking. The company has also made considerable progress on defending against prompt injection attacks. Claude Sonnet 4.5 is released under Anthropic’s AI Safety Level 3 protections.

Claude Sonnet 4.5: Availability

This model is available everywhere. If you are a developer, simply use claude-sonnet-4-5 via the Claude API. Pricing remains the same as Claude Sonnet 4, at $3/$15 per million tokens.

Other than this latest model, the company has also released a set of major upgrades to its products.

Anthropic releases major upgrades for its products

In Claude Code, checkpoints are added that save your progress and allow you to roll back instantly to a previous state. The terminal interface is refreshed, and a native VS Code extension is shipped. A new context editing feature and memory tool have been added to the Claude API that lets agents run even longer and handle even greater complexity.

In the Claude apps, code execution and file creation are brought directly into the conversation. Claude for Chrome extension is made available to Max users who joined the waitlist last month.

Claude Agent SDK is made available to users. It is the same infrastructure that powers Claude Code, but it shows impressive benefits for a wide variety of tasks, not just coding. You can now use the Claude Agent SDK to build your own agents.

The company is also releasing a temporary research preview alongside Claude Sonnet 4.5 called Imagine with Claude. In this experiment, Claude generates software on the fly; no functionality is predetermined, and no code is prewritten. Imagine with Claude is available to Max subscribers for the next four days; you can try it on claude.ai/Imagine.

Claude Code updates are available to all users. Claude Developer Platform updates, including the Claude Agent SDK, are available to all developers and Code execution and file creation are available on all paid plans in the Claude apps.