Anthropic has introduced Claude Haiku 4.5, now available to all users globally. This release targets developers, businesses, and users who depend on AI for real-time, low-latency tasks such as chat assistants, customer service agents, and rapid code prototyping. The model is accessible via the Claude API, Claude Code, and is integrated into platforms like Amazon Bedrock and Google Cloud’s Vertex AI. Pricing is set at $1 per million input tokens and $5 per million output tokens, making it a cost-conscious option for high-volume users.
Introducing Claude Haiku 4.5.
— Alex Albert (@alexalbert__) October 15, 2025
Our latest small model that matches Sonnet 4's performance at a third of the cost and more than twice the speed. pic.twitter.com/JREePxuKSU
Claude Haiku 4.5 delivers coding quality on par with more advanced models like Claude Sonnet 4.5, but with more than double the speed and at a third of the previous cost. The model achieves approximately 90% of Sonnet 4.5's agentic coding performance and outpaces Sonnet 4.5 in responsiveness, especially for sub-agent orchestration and instruction-following tasks. Early user feedback highlights its capacity to handle complex workflows and real-time feedback loops, with some reporting up to five times faster output compared to earlier models. The model also demonstrates marked safety improvements, scoring lower on misaligned behaviors and earning an AI Safety Level 2 (ASL-2) designation, less restrictive than its frontier counterparts.
Anthropic, the company behind Claude, has consistently focused on advancing AI safety and scalability. With the release of Haiku 4.5, Anthropic aims to offer a model that balances intelligence, speed, and cost for both individual developers and enterprise clients, reinforcing its position in the competitive AI landscape.
