{"id":15167,"date":"2025-10-01T21:21:53","date_gmt":"2025-10-01T19:21:53","guid":{"rendered":"https:\/\/haimagazine.com\/uncategorized\/anthropic-presents-claude-sonnet-4-5\/"},"modified":"2025-10-07T10:38:01","modified_gmt":"2025-10-07T08:38:01","slug":"anthropic-presents-claude-sonnet-4-5","status":"publish","type":"post","link":"https:\/\/haimagazine.com\/en\/ai-news-2\/anthropic-presents-claude-sonnet-4-5\/","title":{"rendered":"\ud83d\udd12 Anthropic presents Claude Sonnet 4.5"},"content":{"rendered":"

Anthropic has unveiled its latest model \u2014 Claude Sonnet 4.5. According to the announcements, it’s set to establish a new standard in programming tasks, moving away from the assistant role towards an autonomous tool capable of creating ready-to-deploy applications. The launch comes at a time of increasing competition, especially from OpenAI, and responds to the growing demands of programmers and tech companies.<\/p>

Performance in practice. What can Sonnet 4.5 do?<\/strong><\/h4>

Anthropic’s main argument is sheer performance. The company reports that Sonnet 4.5 delivers top results in the SWE-Bench Verified benchmark, which measures an AI’s ability to tackle real-world programming problems.<\/p>

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

Company representatives, like researcher David Hershey, stress that the tests don’t fully capture the model’s capabilities. During internal trials, Sonnet 4.5 was supposed to autonomously work on a single complex task for over 30 hours. During this time, it not only wrote the code for an application but also set up the necessary databases, purchased a domain, and conducted a SOC 2 security audit.<\/p>

These capabilities have a direct impact on business. The first partners who had access to the model confirm its potential. Cursor SEO says that Sonnet 4.5 represents “state-of-the-art coding performance, specifically on longer horizon tasks,” while Windsurf CEO Jeff Wang shares a similar view, calling it “the new generation of coding models”.<\/p>

Broadening horizons: not just for developers<\/strong><\/h4>

The model’s capabilities aren’t limited to code. Anthropic points out significant progress in logical reasoning and mathematics. Moreover, early tests involving experts from fields like finance, law, medicine and STEM show that Sonnet 4.5 has much better domain knowledge and analytical abilities compared to previous versions. This signals the company’s ambition to create a versatile tool for use across different professional sectors.<\/p>

An ecosystem of tools for developers<\/strong><\/h4>

With the launch of the model, Anthropic is releasing a toolkit designed to make it easier to create advanced solutions. Claude Agent SDK<\/strong>, a set for building your own autonomous AI agents, is at the core of it. The company assures that this is the same infrastructure powering their flagship products.<\/p>

Other services have also been updated. In the Claude Code environment, a checkpoint feature was introduced, allowing users to save their work state and revert changes. There’s also an official extension for the VS Code editor, as well as an option to create office files \u2014 spreadsheets, documents and presentations \u2014 directly in the chat window.<\/p>

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

Anthropic emphasizes that Sonnet 4.5 is their “best-tuned” model in terms of safety. The company claims they’ve greatly reduced undesirable behaviors such as generating false information, the tendency to cheat, or succumbing to harmful commands. They’ve also enhanced the resistance to “prompt injection” attacks, which is very important in the context of creating safe applications. The data below show how the model compares to competitors in terms of undesirable behaviors.<\/p>

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

The model operates under corporate ASL-3 security standards (AI Safety Level 3). This means its built-in filters try to block attempts to generate dangerous content, e.g., related to weapons of mass destruction. Anthropic acknowledges the system isn’t perfect and can mistakenly flag safe queries, but assures that the number of false alarms has been reduced tenfold since the introduction of this technology.<\/p>

Summing up, the launch of Claude Sonnet 4.5 is a strategic move by Anthropic in the fierce tech race. The company is not only showcasing a tool with impressive tested capabilities but it’s also trying to build an entire ecosystem around it, while loudly communicating its approach to safety. Keeping prices at the current level might attract developers and companies looking for an efficient, yet cost-predictable alternative. The next few months will show if Sonnet 4.5 will indeed become the go-to choice for IT professionals.<\/p>

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

Anthropic has introduced Claude Sonnet 4.5, an AI model designed not just to write code, but to create entire applications. The company’s boasting record-breaking results and rolling out new tools for developers.<\/p>\n","protected":false},"author":230,"featured_media":14428,"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-15167","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\/15167","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=15167"}],"version-history":[{"count":1,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts\/15167\/revisions"}],"predecessor-version":[{"id":15168,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts\/15167\/revisions\/15168"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/media\/14428"}],"wp:attachment":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/media?parent=15167"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/categories?post=15167"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/tags?post=15167"},{"taxonomy":"popular","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/popular?post=15167"},{"taxonomy":"difficulty-level","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/difficulty-level?post=15167"},{"taxonomy":"author","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/ppma_author?post=15167"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}