Claude Haiku 4.5 (latest)

Claude Haiku 4.5 occupies Anthropic's fast lane within the Claude family, positioned as a lightweight option tuned for responsive chat, lightweight agent workflows, and routine office tasks. It pairs a broad input surface—accepting text, image, and PDF material—with text-only output, so it fits naturally into pipelines that need to ingest mixed documents and return concise written responses. The model carries the reasoning, tool calling, and temperature controls expected of recent Claude releases, alongside attachment handling, which makes it practical for assistant-style applications that combine document review with structured actions.

On the technical side, the model offers a 200,000-token context window with a 64,000-token maximum output, giving applications room to hold long documents or extended conversation history while still producing substantial replies. Anthropic pricing for direct access is set at roughly one dollar per million input tokens and five dollars per million output tokens, with a discounted cache-read tier that supports reuse of repeated prompts. It is distributed as a closed-weight hosted model rather than an open release, and is available across a wide range of third-party providers that resell the same Anthropic endpoint, which gives teams flexibility in choosing where the inference runs while keeping the underlying behavior consistent.