Anthropic Releases Claude Sonnet 4.5 for Coding
Originally published in Turkish. Read the original →
Anthropic's new model Claude Sonnet 4.5 stands out in coding and intelligent agent development with a 77.2% score on SWE-bench, advanced tool orchestration and long-running task performance; it is available in ecosystems such as GitHub Copilot, Vercel AI Gateway, Amazon Bedrock and Vertex AI.
Anthropic has announced its newest AI model, Claude Sonnet 4.5. With a strong 77.2% score on SWE-bench, a software engineering benchmark, the model has joined the best in its class at boosting coding capabilities and building intelligent agents. Claude Sonnet 4.5 is available through platforms such as Vercel AI Gateway, Amazon Bedrock, Vertex AI and GitHub Copilot.
The model can be used in many IDEs, most notably Visual Studio, JetBrains, Xcode and Eclipse. For enterprise teams, it comes with improvements such as tool orchestration, context editing and domain/industry-specific capabilities, and when paired with solutions like GitLab Duo and Databricks, it aims to increase productivity. It also supports multiple autonomous development workflows, including Next.js builds and frontend design tasks, through Vercel's open-source Coding Agent Platform.
Agentic Capabilities and Long-Horizon Tasks
Claude Sonnet 4.5 delivers significant improvements in agentic capabilities and long-horizon task performance. The model stands out with its ability to sustain tasks lasting more than 30 hours, along with improved memory management and context handling. This makes it possible to run complex processes such as proactive vulnerability patching and advanced financial analysis more reliably in fields like cybersecurity, finance and research. Improvements on the safety and alignment side aim to produce safe and consistent results on complex tasks.
Highlights for Developers and Enterprises
- Tool orchestration: Using multiple tools in a coordinated way and executing tasks by breaking them down.
- Context editing: Better context management across long documents and multi-step processes.
- Domain/industry-specific capabilities: More accurate and compliant outputs on industry-specific tasks.
- Enterprise platforms: Strong use cases together with tools like GitLab Duo and Databricks.
- Autonomous workflows: End-to-end support for processes such as Next.js builds and frontend design; integration with Vercel Coding Agent Platform.
Access via API and SDK
Businesses can tap into the power of Claude Sonnet 4.5 through the Claude API and the Claude Agent SDK. These tools make it easier to build custom agents and create intelligent systems. Developers can more easily reach goals such as smarter context management, reducing prompt injection risks and maintaining secure development environments.
Conclusion
By simplifying the path from idea to workable solution, Claude Sonnet 4.5 cements its leading position among coding-focused AI models. With its capacity to handle complex tasks, its endurance in long-running processes and its progress in enterprise security/compliance, the model adds another strategic step to Anthropic's vision of trustworthy virtual collaborators.
