Anthropic Launches Cheaper AI Model Claude Haiku 4.5
In a move that signals the rapid commoditization of high-level artificial intelligence capabilities, Anthropic has released Claude Haiku 4.5.
The San Francisco-based AI startup describes Haiku 4.5 as its “fastest and most efficient small model yet,” offering near-frontier intelligence at a fraction of the cost.
According to Anthropic’s announcement, the new model delivers “similar levels of coding performance as Claude Sonnet 4, but at one-third the cost and more than twice the speed.”
It’s designed to handle real-time applications — from chat assistants to customer service bots and pair programming — where speed and responsiveness matter most.
Performance and benchmarks
Haiku 4.5’s benchmark results place it in the same league as much larger systems. The model scored 73% on SWE-Bench Verified and 41% on Terminal-Bench, closely matching the performance of Sonnet 4, OpenAI’s GPT-5 and Google’s Gemini 2.5 in coding and command-line tasks.
The model also performs strongly in “computer use” benchmarks, where it outperforms Claude Sonnet 4, allowing smoother interactions in tools like Claude for Chrome.
“It punches way above its weight,” Anthropic’s Chief Product Officer Mike Krieger told CNBC, highlighting how Haiku’s speed and cost efficiency could redefine how AI is deployed at scale.
Free for all, safer than ever
In a highly competitive market, Anthropic is making a bold play for users by making Haiku 4.5 available to everyone on its free plan. This instantly provides the public with access to a level of AI performance that was previously locked behind a paywall.
For developers, the model can be accessed via Claude’s API, Amazon Bedrock, and Google Cloud Vertex AI. Pricing is set at $1 per million input tokens and $5 per million output tokens, the lowest yet for any Claude model.
Alongside performance, the company is emphasizing safety. According to their internal testing, they’ve labeled Haiku 4.5 their “safest model yet.” It showed lower rates of misaligned behavior than its more powerful counterparts and was deemed to pose only limited risks in areas like chemical or biological weapons information. As a result, it has been released under a less restrictive safety classification, the AI Safety Level 2 (ASL-2) standard.
The launch of Haiku 4.5 marks Anthropic’s third major release in two months — following Sonnet 4.5 in late September and Opus 4.1 in August.
Anthropic’s growth trajectory remains strong. The company told Reuters its annual revenue run rate is approaching $7 billion this month, and it now serves more than 300,000 business customers. Last month, Anthropic also announced the completion of a $13 billion Series F fundraising round led by ICONIQ, giving the company a post-money valuation of $183 billion.
Our sister publication, eWeek, recently ranked Anthropic among the top eight generative AI companies and named it one of its AI Visionaries for 2025 — recognizing the company’s position as a leader in generative AI and its ongoing commitment to constitutional AI.
