Claude Sonnet 4.5

Model Overview

Claude Sonnet 4.5 is a major mid-tier model released by Anthropic on September 29, 2025. It was specifically optimized for coding, autonomous agentic workflows, and complex reasoning, featuring a balanced reasoning and cost profile. It shipped with a standard 200K token context window (with up to 1 million tokens available in beta). Note: As of July 2026, it has been largely superseded by the newer Claude Sonnet 5.

Capabilities

• Extended Thinking Mode: Allows for more deliberate, step-by-step problem-solving before generating outputs.
• Coding Proficiency: Highly regarded for its coding abilities, designed for long-horizon tasks capable of maintaining focus and continuity for 30+ hours on complex projects.
• Balanced Reasoning: Provides a "Goldilocks" profile that blends high speed, low cost, and strong reasoning capabilities.

Example Use Cases

• Complex, multi-step software development and autonomous coding agents.
• Deep data analysis and logical problem-solving requiring extended thinking.
• Enterprise workflows requiring robust tool use and API orchestration.
• Processing massive codebases or large datasets using the extended context window.

Performance & Benchmarks

At launch, Claude Sonnet 4.5 achieved a remarkable 82% on the SWE-bench Verified benchmark, showcasing its state-of-the-art capability in software engineering tasks. Its extended thinking mode allowed it to rival larger, more expensive models on complex logic puzzles.

Intended Use & Limitations

Intended for developers and enterprises needing powerful, long-horizon agentic capabilities without the premium cost of the Opus/Fable tiers. While highly capable, it is no longer the latest model in the Sonnet family, and users deploying new applications are generally directed towards Claude Sonnet 5.

About Anthropic

Anthropic is an AI safety and research company focused on building reliable, interpretable, and steerable AI systems. They are the creators of the Claude family of models, known for prioritizing safety and alignment alongside frontier capability.