Claude Haiku 4.5

Claude Haiku 4.5 is the fast, economical member of Anthropic's Claude 4.5 family — a closed-weights model aimed at high-volume, latency-sensitive work: coding agents, sub-agent orchestration, classification, routing, extraction, and customer-service automation. It is the first Haiku model with reasoning support, accepts text and image input, offers a 200K-token context window with 64K max output, and has a February 2025 knowledge cutoff. Anthropic positions it as a drop-in replacement for both Haiku 3.5 and Sonnet 4 at its most economical price point, and rates it ASL-2 (safer classification than Sonnet 4.5/Opus 4.1's ASL-3).

Provider-reported: Anthropic claims coding performance similar to Sonnet 4 at one-third the cost and 2x+ the speed, and that Haiku 4.5 surpasses Sonnet 4 on computer-use tasks (SWE-bench Verified comparison chart on the announcement page). Partner testimonials: Augment reports 90% of Sonnet 4.5's agentic coding evaluation performance; GitHub Copilot reports comparable quality to Sonnet 4 at faster speed; Warp, Gamma and others cite instruction-following and responsiveness gains. Independent: launch covered by independent commentators (e.g. Simon Willison's analysis confirming specs); independent third-party benchmark reproductions were still thin at launch — treat performance claims as provider-claimed.