Claude Haiku 4.5: the small model that matches May's Sonnet 4
In one sentence Anthropic releases Claude Haiku 4.5: performance equal to Claude Sonnet 4 (May 2025) at a third of the price and double the speed. Changes the cost/quality ratio for high-volume agentic tasks.
Anthropic completes the Claude 4 family with Haiku 4.5 — the small, fast, cheap model. The impressive part: it has performance similar to May 2025's Sonnet 4 (the "medium" model from just 5 months ago).
Translation: in five months, what was Anthropic's "workhorse" model has become the "small model". Same quality level, but at:
- $1 input / $5 output per million tokens (Sonnet was $3/$15);
- more than double the speed;
- well-suited to repetitive tasks (email classification, PDF data extraction, agents making many short calls).
For AI product builders: Haiku 4.5 is a "compute budget" inflection. Operations that cost $X in May now cost $X/3, allowing the jump from prototype to production at high volume.
Companies
Anthropic
Tools
Claude Haiku 4.5
Tags
Sources