Claude Sonnet 4.5

llm

Anthropic's coding-focused model that set a new state of the art on SWE-bench Verified at launch. The 82.0% figure reported with additional test-time compute is deliberately not recorded.

Metadata

org	Anthropic
released	2025-09-29
license	proprietary
kind	llm

Blog ↗

Benchmarks

SWE-bench

77.2

source ↗

Lineage

No recorded lineage — root or standalone model.

view in full graph →