Anthropic Releases Claude Sonnet 4.5 With Long‑Horizon Coding
The new model posts a 61.4 percent OSWorld score and sustains autonomous work for more than 30 hours on complex coding tasks.
What Anthropic announced on September 29
Anthropic introduced Claude Sonnet 4.5 and an upgraded Opus 4.1, positioning both as improvements for complex, multi‑step software work. The company highlighted stronger focus, longer task persistence, and reduced harmful behaviors. Pricing for Sonnet 4.5 remains $3 per million input tokens and $15 per million output tokens, keeping it a drop‑in replacement.
Benchmarks and model behavior
Anthropic cited a 61.4 percent result on the OSWorld benchmark that evaluates real‑world computer tasks. The company says Sonnet 4.5 can run autonomously for 30 plus hours on intricate projects, compared with seven hours for Claude …
Archive Access
This article is older than 24 hours. Create a free account to access our 7-day archive.