Claude Sonnet 4.5 launch and positioning
The next generation of Claude
Anthropic has released Claude Sonnet 4.5 and is calling it “the best coding model in the world,” “the strongest model for building complex agents,” and “the best model at using computers.” The upgrade lands alongside improvements across popular Claude products and is available immediately, with pricing unchanged from Sonnet 4.
Clear progress over its predecessor
OpenAI’s latest study found Claude Sonnet 4.1 outperformed GPT-5, Gemini, and Grok on real-world job tasks. Building on that, Anthropic says Sonnet 4.5 delivers “clear progress,” achieving higher performance in benchmarks and better meeting practical business needs, enabling customers to “do more, solve harder problems, and be more creative.”
Agentic capabilities and autonomous operation
Natural-language agent creation
Sonnet 4.5 can help users build custom agents using natural language, simplifying complex automation and orchestration tasks that previously required specialized tooling or extensive configuration.
Long-horizon autonomy
Anthropic claims Sonnet 4.5 can work autonomously for 30 hours, a major jump from Claude Opus 4’s 7-hour capability. This extended focus is designed for complex, multi-step workflows where consistency and incremental progress matter.
Coding upgrades and Claude Code checkpoints
Checkpoints for reliable iteration
Claude Code now includes checkpoints, a highly requested feature that lets you save progress and roll back to previous states instantly. This reduces risk during refactors, experiments, and long-running coding sessions.
Practical improvements for developers
Alongside checkpoints, Claude’s coding experience benefits from code execution and streamlined workflows across documents and structured assets, helping teams move faster from idea to implementation.
In-app code execution and file creation
Create spreadsheets, slides, and docs inside Claude
Within the Claude apps, you can now execute code and generate files—spreadsheets, slides, and documents—directly in conversation. These creation capabilities are available on all paid plans, improving end-to-end productivity without context switching.
Expanded access via Chrome
Claude for Google Chrome will become available to everyone who previously joined the waitlist, bringing upgraded computer-use capabilities into the browser for task completion and workflow support.
Alignment and safety improvements
The most aligned Claude model yet
Anthropic says Sonnet 4.5 is its “most aligned model yet,” substantially reducing behaviors like sycophancy, deception, power-seeking, and tendencies to encourage delusional thinking. This aims to deliver more trustworthy assistance in everyday and high-stakes scenarios.
Safer agentic and computer-use workflows
The company highlights progress in defending against prompt injection attacks—one of the most serious risks for AI systems that use tools and computers—so users can deploy agentic capabilities with stronger safeguards.
Availability, pricing, and access
Available today with unchanged pricing
Claude Sonnet 4.5 is available today, and prices remain the same per million tokens as Sonnet 4. That parity makes upgrading straightforward for teams already using Claude.
Paid-plan features for teams
Capabilities such as code execution and file creation in the Claude apps are available on all paid plans, encouraging broader adoption across business functions without additional licensing friction.
Early customer reactions and business impact
Canva’s take on intelligence and scale
Danny Wu, Head of AI Products at Canva, reports “impressive gains on our most complex, long-context tasks—from engineers in our codebase to in-product features and research,” noting Sonnet 4.5 is “noticeably more intelligent” and helps push what 240M+ users can design.
Enterprise autonomy for large codebases
Sean Ward, CEO of iGent AI, says Sonnet 4.5 “resets our expectations—it handles 30+ hours of autonomous coding,” enabling engineers to tackle complex architectural work faster while maintaining coherence across massive codebases.
FAQs
What makes Claude Sonnet 4.5 stand out?
It’s positioned as the best model for real-world agents, coding, and computer use, with long-horizon autonomy (up to 30 hours), improved alignment, and practical upgrades like checkpoints in Claude Code and in-app file creation.
Is Claude Sonnet 4.5 more expensive than previous versions?
No. Pricing remains the same per million tokens as Claude Sonnet 4, simplifying upgrades for current users.
Which features are available on paid plans?
You can run code and create files—spreadsheets, slides, and documents—directly in Claude conversations on all paid plans, streamlining end-to-end workflows.
