Smaller, Faster, And Cheaper: Claude Haiku 4.5 AI Model Is The New Default For Claude Free Users

Anthropic released a new small model on Wednesday that might have been spotted in some leaks recently: Claude Haiku 4.5. The AI firm says the small model will offer the same coding performance as Claude Sonnet 4, a state-of-the-art AI model Anthropic launched in May. A few weeks ago, Anthropic released Claude Sonnet 4.5, its newest flagship AI product. Claude Haiku 4.5 will become the new default model for Claude subscribers who are on the Free plan.

Anthropic said in a blog post that Claude Haiku 4.5 is more than two times faster than Sonnet 4 at producing similar levels of coding performance, while the costs have dropped to one-third of Sonnet 4's costs. Benchmark results that Anthropic provided show that Claude Haiku 4.5 outperformed ChatGPT (GPT-5) and Gemini 2.5 Pro in addition to Sonnet 4 in software engineering tests. Claude Haiku 4.5 also had high scores in other benchmark tests, including agentic tool use, computer use, visual reasoning, and others, as seen below.

Anthropic

When it comes to safety, Anthropic says that Claude Haiku 4.5 showed a "statistically significantly lower overall rate of misaligned behaviors" than Sonnet 4.5 and Opus 4.1." Thus, Claude Haiku 4.5 is Anthropic's safest model yet. Claude Haiku 4.5 will be available everywhere on Wednesday, with pricing set at $1/$5 per million input and output tokens.

The Claude Haiku 4.5 benchmarks Anthropic shared are promising. At the very least, Claude users who are on the Free subscription should get a bump in performance and a faster experience. But the smaller Claude model isn't just a solution Anthropic proposes to Claude users who want access to a reliable chatbot without paying a monthly subscription. Anthropic says the Claude Haiku 4.5 might be the AI model that best meets the needs of certain types of consumers. "Users who rely on AI for real-time, low-latency tasks like chat assistants, customer service agents, or pair programming will appreciate Haiku 4.5's combination of high intelligence and remarkable speed," Anthropic notes.

Also, Claude Haiku 4.5 gives developers new ways of using multiple Claude models together. One example involves Sonnet 4.5 breaking down a complex problem into multi-step plans. A team of multiple Claude Haiku 4.5 models would complete the subtasks at the same time.

Some enterprise users have already tested Claude Haiku 4.5, praising the new AI model, according to comments Anthropic provided. "Speed is the new frontier for AI agents operating in feedback loops," Block Inc's Brad Axen said in a statement. "Haiku 4.5 proves you can have both intelligence and rapid output. It handles complex workflows reliably, self-corrects in real-time, and maintains momentum without latency overhead. For most development tasks, it's the ideal performance balance." Similarly, Warp AI CEO Zach Lloyd said that "Haiku 4.5 is a leap forward for agentic coding, particularly for sub-agent orchestration and computer use tasks. The responsiveness makes AI-assisted development in Warp feel instantaneous."