claude-sonnet-4-5-20250929 · claude_lite

Claude Sonnet 4.5 is Anthropic's most advanced Sonnet model to date, optimized for real-world agents and programming workflows. It delivers outstanding performance in programming benchmarks such as SWE-bench Verified, with improvements in system design, code security and compliance. Designed for long-duration autonomous operation, the model maintains task continuity across sessions and enables fact-based progress tracking. Sonnet 4.5 also boasts upgraded agent capabilities, including enhanced tool orchestration, speculative parallel execution, and more efficient context and memory management. Equipped with advanced context tracking and token utilization awareness across tool calls, it is well-suited for multi-context and long-running workflows. Its application scenarios span software engineering, cybersecurity, financial analysis, research agents and other fields requiring sustained reasoning and tool utilization.