Anthropic introduces Opus 4.5, cuts API pricing, and enables much longer Claude chats

Anthropic has officially launched Opus 4.5, its latest flagship model, showcasing significant advancements in coding efficiency and user experience that position it as a strong competitor to OpenAI's frontier models. One of the standout features of this update is the enhanced ability for Claude to maintain longer conversations without abruptly terminating them. Previously, users faced interruptions in their chats when they reached the hard context limit of 200,000 tokens. Instead of continuing the conversation in a fragmented manner, Claude would end the interaction, which left many users frustrated. However, the new Opus 4.5 model introduces a more sophisticated memory management system. It summarizes key points from earlier parts of the conversation, allowing for a more coherent dialogue even as it approaches the maximum token limit. In addition to improving user experience, Opus 4.5 is also the first model to achieve an accuracy score exceeding 80 percent, specifically reaching 80.9 percent on the SWE-Bench Verified benchmark. This puts it ahead of OpenAI's GPT-5.1-Codex-Max, which scored 77.9 percent, and Google’s Gemini 3 Pro at 76.2 percent. While Opus 4.5 excels in coding tasks and tool use scenarios, it still trails behind GPT-5.1 in visual reasoning capabilities. Developers using Anthropic’s API will also benefit from these enhancements, as they can implement similar context management features in their applications. With these updates, Anthropic aims to redefine user interactions with AI by prioritizing smoother and more meaningful conversations.