{"id":66807,"date":"2026-04-16T19:15:54","date_gmt":"2026-04-16T17:15:54","guid":{"rendered":"https:\/\/www.apfelpatient.de\/?p=66807"},"modified":"2026-04-16T19:17:29","modified_gmt":"2026-04-16T17:17:29","slug":"claude-opus-4-7-anthropics-new-ai-model-is-here","status":"publish","type":"post","link":"https:\/\/www.apfelpatient.de\/en\/news\/claude-opus-4-7-anthropics-neues-ki-modell-ist-da","title":{"rendered":"Anthropic introduces Claude Opus 4.7: Focus on sophisticated Software Development"},"content":{"rendered":"

Anthropic has released its latest AI model, Claude Opus 4.7. The update brings significant improvements to complex coding tasks, better image processing, and an optimized saving function for multi-part projects.<\/strong><\/p>\n\n\n\n

Anthropic is consistently maintaining its two-month cycle for model updates: Following Opus 4.5 in November and 4.6 in February, Opus 4.7 will be released in April. The company \u2013 which is collaborating<\/a> with Apple on cybersecurity as part of Project Glasswing \u2013 is positioning the new model as a direct upgrade with a focus on advanced software development.<\/p>\n\n\n\n

Claude Opus 4.7 replaces<\/a> Opus 4.6 as Anthropic's most powerful, generally available model. According to Anthropic, the model shows the greatest improvements in the most difficult tasks\u2014precisely those coding problems that required close developer oversight with previous models. Opus 4.7 is designed to handle complex, long-running tasks more reliably, follow instructions more accurately, and verify its own results before reporting them back.<\/p>\n\n\n\n

What makes Opus 4.7 better<\/h2>\n\n\n\n

The main improvements cover several areas. In software development, the focus is on agent-based coding \u2013 that is, the autonomous handling of complex, multi-stage tasks. According to Anthropic, users report that they can delegate their most demanding programming tasks to Opus 4.7 without having to constantly monitor them.<\/p>\n\n\n\n

The model also features significantly improved image processing: it can analyze higher-resolution images. For professional tasks such as creating user interfaces, presentations, and documents, the model delivers higher-quality results.<\/p>\n\n\n\n

In benchmarks, Opus 4.7 outperforms its predecessor Opus 4.6, OpenAI's GPT-5.4, and Google's Gemini 3.1 Pro in agent-based coding and computer use. Only Claude Mythos Preview \u2013 the specialized cybersecurity model available exclusively to partners like Apple \u2013 performs better across a broader range.<\/p>\n\n\n\n

Improved memory and token changes<\/h2>\n\n\n\n

Another improvement concerns file-based memory: Opus 4.7 retains important information across long, multi-part work sessions and requires less context from the outset for subsequent tasks. For developers who work on a project for days, this represents a significant productivity gain.<\/p>\n\n\n\n

Two changes API users should be aware of during migration: A new tokenizer can cause the same input to consume 1.0 to 1.35 times as many tokens as before. Additionally, Opus 4.7 performs more detailed reasoning at higher complexity levels, which improves reliability for difficult problems but also produces more output tokens.<\/p>\n\n\n\n

Claude Code: New features for Mac users<\/h2>\n\n\n\n

Alongside the new model, Anthropic is also expanding<\/a> Claude Code for Mac. The Auto Mode, which offers developers a safer alternative to manually confirming each step, is now available to Max plan subscribers. Also new is the "\/ultrareview" command, which starts a dedicated review session and checks changes as a meticulous human reviewer would.<\/p>\n\n\n\n

Anthropic's two-month update cycle signals that the company has no intention of slowing down its model development. Users of the Claude desktop app or API can expect regular, substantial improvements. (Image: Shutterstock \/ Stockinq)<\/p>\n\n\n\n