← Back to glossary
+Suggest a term
Concept·AI Models & Capabilities·Added 1 day ago

Claude Haiku

Also known as: Haiku

The fast, low-cost tier in Anthropic's Claude family, built for high-volume, latency-sensitive tasks. As of May 2026, Claude Haiku 4.5 is the current version: priced at $1 input / $5 output per million tokens and suited for real-time chat, bulk processing, classification, and subtask execution within multi-agent workflows.

Haiku is the lightweight end of Anthropic's three-tier model lineup. The name reflects the design goal: fast, compact, and efficient. Haiku 3 launched alongside Claude 3 in March 2024, and Claude 3.5 Haiku followed in late 2024 with improved performance. The current version is Claude Haiku 4.5, released October 2025, which targets teams that need low-latency responses at high volume without paying Sonnet or Opus prices.

Haiku 4.5 retains capabilities closer to the frontier than earlier small models, making it meaningfully different from a throwaway routing tier. Anthropic positions it for real-time customer support bots, short-text classification, quick code edits, and as a fast worker executing subtasks within larger multi-agent orchestrations where a heavyweight model would be wasteful. The $1/$5 pricing makes it practical to route large volumes of simple queries without the inference costs adding up to a budget concern.

One practical note: Haiku has not received minor-version updates at the same cadence as Sonnet and Opus. As of May 2026, Haiku 4.0, 4.6, 4.7, and 4.8 do not exist. When building multi-model routing systems, the safe pattern is to default to Sonnet for most requests and fall back to Haiku for tasks where context is simple and throughput matters more than depth.

This definition is AI-generated and refreshed weekly. It may contain inaccuracies. Use your own judgment, especially for production decisions.
Related terms
ClaudeClaude SonnetClaude OpusAnthropicInference cost