{"id":15692,"date":"2025-10-17T14:45:18","date_gmt":"2025-10-17T12:45:18","guid":{"rendered":"https:\/\/haimagazine.com\/uncategorized\/claude-haiku-4-5-more-power-at-a-lower-price\/"},"modified":"2025-10-24T09:25:20","modified_gmt":"2025-10-24T07:25:20","slug":"claude-haiku-4-5-more-power-at-a-lower-price","status":"publish","type":"post","link":"https:\/\/haimagazine.com\/en\/ai-news-2\/claude-haiku-4-5-more-power-at-a-lower-price\/","title":{"rendered":"\ud83d\udd12 Claude Haiku 4.5: more power at a lower price"},"content":{"rendered":"

Anthropic is expanding its family of language models with the introduction of Claude Haiku 4.5. This latest compact model combines high performance and advanced capabilities at a low cost. Its launch makes sophisticated AI technology more accessible for everyday use.<\/p>

Performance and savings<\/strong><\/h4>

Claude Haiku 4.5 matches Claude Sonnet 4 in programming performance, and the latter was just launched five months ago. It\u2019s more than twice as fast and available at one third of the price. Moreover, in some tasks like computer handling, the new model even outperforms its predecessor.<\/p>

How does Haiku 4.5 perform in tests?<\/strong><\/h4>

Comparative data confirm the great potential of the new model. In the SWE-bench Verified software engineering test, Haiku 4.5 scored 73.3%, not only surpassing Sonnet 4 (72.7%) but also outperforming models like GPT-5 (72.8%) and Gemini 2.5 Pro (67.2%).<\/p>

\"\"
Source: Anthropic<\/figcaption><\/figure>

Detailed tests show that Haiku 4.5 outperforms Sonnet 4 in many key areas. Besides the mentioned software engineering, it handles better computer management tasks (OSWorld benchmark \u2013 scoring 50.7% compared to 42.2%), terminal programming (Terminal-Bench \u2013 41.0% to 36.4%), and solving high school math problems (AIME 2025 \u2013 96.3% to 70.5% using Python).<\/p>

\"\"
Source: Anthropic<\/figcaption><\/figure>

Thanks to these improvements, apps like the Claude for Chrome extension operate more smoothly and become more useful. People who use AI for tasks that require an immediate response, such as in chat assistance, customer service, or pair programming, will particularly appreciate the combination of high intelligence and fast speed in Haiku 4.5. Meanwhile, in the Claude Code environment, this model significantly enhances responsiveness \u2014 from multi-agent projects to rapid prototype development.<\/p>

New opportunities in model collaboration<\/strong><\/h4>

Claude Sonnet 4.5, unveiled two weeks ago, remains the most advanced model in Anthropic’s lineup. Haiku 4.5 offers an alternative aimed at those seeking high performance at significantly lower costs.<\/p>

The new model also opens up additional ways to use technology. For example, Sonnet 4.5 can break down a complex problem into multi-stage plans, and then coordinate the work of a team made up of several Haiku 4.5 instances, which work on smaller tasks simultaneously.<\/p>

Safety and availability<\/strong><\/h4>

The Claude Haiku 4.5 model has undergone thorough safety assessments. It showed a low rate of undesirable behaviors and is much better aligned with guidelines than its predecessor, Haiku 3.5. In automated compliance tests, it achieved a score that indicates a lower overall rate of improper behaviors compared to Sonnet 4.5 and Opus 4.1.<\/p>

Tests also showed that the model presents a limited risk in the context of generating content related to chemical, biological, radiological, and nuclear (CBRN) weapons. For this reason, it has been classified at the AI Safety Level 2 (ASL-2), which is less restrictive than the ASL-3 level assigned to larger models.<\/p>

<\/p>","protected":false},"excerpt":{"rendered":"

The new Claude Haiku 4.5 model delivers performance similar to the best solutions, but it’s much faster and cheaper. Check out how it performs in tests.<\/p>\n","protected":false},"author":230,"featured_media":13126,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"rank_math_lock_modified_date":false,"footnotes":""},"categories":[813],"tags":[707,887,817],"popular":[],"difficulty-level":[36],"ppma_author":[884],"class_list":["post-15692","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai-news-2","tag-ai-5","tag-anthropic-2","tag-claude-2","difficulty-level-easy"],"acf":[],"authors":[{"term_id":884,"user_id":230,"is_guest":0,"slug":"karolina-ceron","display_name":"Karolina Cero\u0144","avatar_url":"https:\/\/haimagazine.com\/wp-content\/uploads\/2025\/07\/PXL_20250419_110132091.MP4-scaled.jpg","first_name":"Karolina","last_name":"Cero\u0144","user_url":"","job_title":"","description":"Wsp\u00f3\u0142tw\u00f3rczyni newslettera AI Flash, studentka psychologii i pasjonatka sztucznej inteligencji. Interesuj\u0119 si\u0119 wp\u0142ywem nowych technologii na cz\u0142owieka, a w wolnych chwilach eksperymentuj\u0119 z generatywn\u0105 grafik\u0105 w Midjourney."}],"_links":{"self":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts\/15692","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/users\/230"}],"replies":[{"embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/comments?post=15692"}],"version-history":[{"count":1,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts\/15692\/revisions"}],"predecessor-version":[{"id":15693,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts\/15692\/revisions\/15693"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/media\/13126"}],"wp:attachment":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/media?parent=15692"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/categories?post=15692"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/tags?post=15692"},{"taxonomy":"popular","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/popular?post=15692"},{"taxonomy":"difficulty-level","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/difficulty-level?post=15692"},{"taxonomy":"author","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/ppma_author?post=15692"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}