Back to products
Claude Haiku 4.5

Claude Haiku 4.5

The fastest, most affordable coding model

Overview

What it is

Anthropic just unveiled Claude Haiku 4.5, its fastest and most efficient small model yet. Haiku 4.5 matches the coding performance of Claude Sonnet 4, once a frontier model, but now runs 2x faster and costs one-third as much. I

Intent

I need it when

Build fast, cost-efficient AI chat assistants and customer service agents

Haiku 4.5 delivers near-frontier coding quality at one-third the cost of Sonnet 4 and more than twice the speed, making it ideal for real-time, low-latency applications like chat assistants and customer service agents that need responsive interactions without high operational costs

Deploy multi-agent systems with orchestrated subtask execution

Haiku 4.5 enables cost-effective parallel processing of subtasks; frontier models like Sonnet 4.5 can break down complex problems into plans and orchestrate teams of multiple Haiku 4.5 instances to complete work in parallel, reducing overall latency and cost

Build responsive AI-powered development tools and IDE integrations

Haiku 4.5's combination of high intelligence and remarkable speed makes it ideal for IDE integrations like GitHub Copilot and development environments where instantaneous responsiveness and real-time feedback loops are essential for user experience

Implement rapid pair programming and code generation workflows

Haiku 4.5 achieves 90% of Sonnet 4.5's coding performance while running 4-5x faster, enabling developers to use it for pair programming, rapid prototyping, and code generation tasks where speed and responsiveness are critical

Reduce API costs while maintaining high-quality model performance

At $1/$5 per million input/output tokens, Haiku 4.5 provides similar coding performance to Claude Sonnet 4 at one-third the cost, allowing developers and organizations to accomplish more within usage budgets while maintaining premium model quality

Drop

Not a fit when

  • User requires frontier-level reasoning for complex multi-step problems; Claude Sonnet 4.5 or Opus 4.8 are better choices
  • Application demands the absolute highest coding performance; Sonnet 4.5 remains the best coding model
  • User needs extended context windows beyond Haiku's capabilities for very long documents
  • Workload requires guaranteed low-latency response times in production with strict SLA requirements not yet published
  • User prioritizes maximum safety guardrails; Claude Haiku 4.5 is released under ASL-2 (less restrictive than ASL-3 models)
Commercials

Pricing

Pay-per-token API pricing; also available through Claude web/desktop plans (Free, Pro, Max, Team, Enterprise) View pricing