Claude Haiku 4.5

Claude Haiku 4.5 extends Anthropic's lightweight model tier with capabilities that previously sat in higher-priced variants. It is described as the first Haiku release to incorporate extended thinking, computer use, and context awareness, signaling a deliberate effort to compress frontier-class behavior into a smaller, faster package. The model fits within a recent Claude release sequence, following Claude Opus 4.1 and Claude Sonnet 4.5, and is positioned by Anthropic as its fastest and most cost-efficient option, matching the performance of its predecessor Sonnet 4 on coding, computer use, and agent-driven tasks.

In practical terms, Claude Haiku 4.5 is aimed at scaled deployments where latency and price matter but coding competence cannot be compromised. Anthropic highlights a SWE-bench Verified score of 73.3%, placing it among the stronger coding models available at its efficiency tier. Third-party testing suggests the model performs close to Sonnet 4.5 at roughly a third of the cost, making it a natural fit for multi-agent orchestration, high-volume coding assistants, and production agent pipelines where near-frontier intelligence needs to be affordable and responsive.