Claude Sonnet 4.5 launches — Anthropic claims it's the world's best coding and agentic model
Sep 29, 2025
Key Points
- Anthropic launches Claude Sonnet 4.5, claiming it surpasses Opus in capability while costing one-fifth as much to operate.
- Sonnet 4.5 targets coding and agent-building as the near-term competitive battleground, positioning it as the strongest model for autonomous workflows.
- The pricing advantage addresses enterprise cost sensitivity at a moment when teams are evaluating which model to embed in production systems.
Summary
Anthropic launched Claude Sonnet 4.5, claiming it surpasses Opus in capability while costing one-fifth as much to run. The company is positioning it as the strongest model for coding and agent-building, a category that includes autonomous coding assistants, task automation, and multi-step reasoning workflows.
OpenAI's o1 and other recent releases have dominated headlines on reasoning and complex tasks. Anthropic is betting that developer velocity and agent-building represent the near-term battleground instead.
The pricing matters at a moment when enterprise and startup buyers are cost-conscious about inference. At one-fifth the operational expense of Opus for equivalent or better performance, Sonnet 4.5 could shift how teams evaluate which model to embed in production systems.