Anthropic has released its latest AI model, Claude Sonnet 4.5. The company claims that Sonnet 4.5 can work autonomously for 30 hours to create apps from scratch. This is a massive improvement to Anthropic’s Opus 4.1 model that could work autonomously for seven hours.
According to Anthropic, Claude Sonnet 4.5 created a chat app similar to Slack and Microsoft Teams with over 11,000 lines of code. The AI model only stopped after the task was completed. The company claims that Sonnet 4.5 is the “best model at using computers,” with a focus on building complex agents.
Claude Sonnet 4.5 expands on the Computer Use feature debuted last year. It allows for the AI model to understand what’s on the screen and navigate according to the task at hand, similar to humans.
Anthropic insists that Sonnet 4.5 is its most aligned frontier model yet, i.e., the company has managed to reduce “concerning behaviours like sycophancy, deception, power-seeking, and the tendency to encourage delusional thinking.”
Anthropic brings new features to Claude Code and Claude API
Anthropic has also brought updates to Claude Code. Now, users get a new checkpoint feature. This will allow you to easily roll back to previous versions of code in any project.
Additionally, the company has updated the Claude API, allowing agents to run for longer and solve more complex problems. You can now even create file documents such as slides, spreadsheets and documents directly during a conversation with Claude.
Anthropic launches Imagine with Claude
AI is all about creativity. Users love to use models such as Google’s Nano Banana to bring their imagination to life. Now, Claude will let you express your imagination through software. Anthropic’s Imagine with Claude feature creates on-the-fly software based on a user’s prompt, with no preset codes or commands.
In a tutorial, the company showed Claude generate a desktop for William Shakespeare from a single prompt, with the AI adding features such as play scripts, show timings and more.
Is Claude Sonnet 4.5 free?
Free users will get access to Claude Sonnet 4.5, though it is likely that it will be restricted to a set number of per-day tokens. Anthropic has not raised prices for its premium members while giving access to its latest AI model.
In a recent study by OpenAI, Claude Opus 4.1 had outperformed all AI agents, including GPT-5, and almost reached parity with humans in selected tasks. Claude Sonnet 4.5 may further improve upon Opus 4.1’s performance.
– Ends