Claude Sonnet 4.5 is Anthropic’s most advanced Sonnet model to date, optimized for real-world agents and coding workflows. It delivers state-of-the-art performance on coding benchmarks such as SWE-bench Verified, with improvements across system design, code security, and specification adherence. The model is designed for extended autonomous operation, maintaining task continuity across sessions and providing fact-based progress tracking.
Sonnet 4.5 also introduces stronger agentic capabilities, including improved tool orchestration, speculative parallel execution, and more efficient context and memory management. With enhanced context tracking and awareness of token usage across tool calls, it is particularly well-suited for multi-context and long-running workflows. Use cases span software engineering, cybersecurity, financial analysis, research agents, and other domains requiring sustained reasoning and tool use.
Reasoning max_tokens set to 2000.
Evaluations
17
across 15 benchmarks
Latency
21.6s
Context length
200k
tokens
Cost
$3 · $15
input · output per 1M tokens
Key takeaways
Claude Sonnet 4.5 is optimized for real-world agents and coding workflows, specifically designed for tasks requiring sustained reasoning and tool use across domains like software engineering, cybersecurity, and financial analysis.
It utilizes a Mixture of Experts (MoE) architecture, supports multimodal input (text+image to text), and features advanced agentic capabilities like tool orchestration and efficient context management.
The model demonstrates high proficiency in core Python programming tasks, achieving 97.56% accuracy on 'Human Evaluation' for Python Standard Library Tasks, solving 58 out of 62 problems.
Known limitations include struggles with ambiguous prompts, nuanced scientific contexts, spatial/probabilistic reasoning, counterfactual questions, and algebraic errors, often exhibiting confabulations and errors in information retrieval.
It scored only 8.11% accuracy on 'Humanity's Last Exam' due to issues with ambiguous prompts, difficulty managing long chains of reasoning, and confabulations, despite showing strengths in well-defined problems.
