Particle.news

Download on the App Store

Anthropic Releases Claude Haiku 4.5, a Faster, Cheaper Small Model for Broad Use

Anthropic pitches an ASL-2 small model for real-time developer workflows.

Overview

  • Claude Haiku 4.5 is available today across Anthropic apps and Claude Code as well as the API, Amazon Bedrock, and Google Cloud Vertex AI, and it becomes the default for free users while serving as the lowest-cost option for paid tiers.
  • Pricing is set at $1 per million input tokens and $5 per million output tokens.
  • Anthropic reports similar coding performance to Sonnet 4 at roughly one-third the cost and more than twice the speed, with strong suitability for low-latency tasks like chat assistants and coding copilots.
  • Internal benchmarks include 73% on SWE-Bench Verified and 41% on Terminal-Bench, placing Haiku 4.5 on par with Sonnet 4, OpenAI’s GPT-5, and Google’s Gemini 2.5 on many tasks while remaining below Sonnet 4.5 overall.
  • Anthropic classifies the model at AI Safety Level 2 and says safety tests show lower rates of misaligned behaviors than Sonnet 4.5 and Opus 4.1, and it is positioned to act as fast sub-agents under Sonnet 4.5 in multi-agent workflows.