Claude Haiku 4.5
Positioning
Claude Haiku 4.5 is a small model designed for low latency and cost efficiency. At launch, Anthropic reported coding performance comparable to Claude Sonnet 4 at one-third the cost and more than twice the speed. The model is now listed as active but legacy. (anthropic.com)
Capabilities and uses
- Accepts text and image inputs and produces text output. It supports manually enabled extended thinking. (platform.claude.com)
- Intended uses include coding, pair programming, rapid prototyping, customer service, chat assistants, computer-use tasks and multi-agent workflows. (anthropic.com)
- The Claude API alias is
claude-haiku-4-5; its pinned snapshot isclaude-haiku-4-5-20251001. Platform-specific identifiers differ. (platform.claude.com)
Limits
- Context window: 200,000 tokens; maximum output: 64,000 tokens.
- Reliable knowledge cutoff: February 2025; training data cutoff: July 2025.
- Adaptive thinking is not supported; extended thinking must be enabled manually. (platform.claude.com)
Pricing
For the first-party Claude API, standard global pricing is USD $1 per million input tokens and USD $5 per million output tokens. Prompt-cache writes cost USD $1.25 per million tokens for a five-minute cache or USD $2 per million tokens for a one-hour cache; cache reads cost USD $0.10 per million tokens. The Batch API provides a 50% discount on input and output token prices. Other platforms and regional endpoints may have different pricing. Official pricing documentation; checked November 17, 2026. (platform.claude.com)
Release and official sources
Released October 15, 2025. See the official announcement and model documentation. (anthropic.com)