🤯 AI Just Leveled Up! Opus 4.5 🚀

Tech & Science

On Monday, Anthropic announced Opus 4.5, the final release in its 4.5 series of models, following the September launch of Sonnet 4.5 and the October release of Haiku 4.5. As anticipated, Opus 4.5 demonstrates state-of-the-art performance across a variety of benchmarks, including coding benchmarks (SWE-Bench and Terminal-bench), tool use (tau2-bench and MCP Atlas), and general problem-solving (ARC-AGI 2, GPQA Diamond). Notably, Opus 4.5 represents the first model to achieve a score over 80 percent on the SWE-Bench verified test. Anthropic underscored the model’s capabilities in computer use and spreadsheet applications, coinciding with the broader release of related products, including Claude for Chrome and Claude for Excel, previously in pilot programs. Claude for Chrome will be available to all Max users, while Claude for Excel will be accessible to Max, Team, and Enterprise users. Furthermore, Opus 4.5 incorporates memory improvements designed for long-context operations, a change that necessitated significant modifications to the model’s memory management. “We’ve made improvements to general long-context quality through training with Opus 4.5, but context windows alone are not sufficient,” Dianne Na Penn, Anthropic’s head of product management for research, stated to TechCrunch. “Knowing the right details to remember is really important in complement to just having a large context window.”

The updated Claude model, Opus 4.5, incorporates a long-requested “endless chat” feature for paid users. This functionality allows conversations to continue uninterrupted, even when the model reaches its context window limit; instead of halting, the model will automatically compress its contextual memory without notifying the user. “This highlights the importance of fundamental elements such as memory,” Penn explains, “because Claude needs to be capable of exploring complex codebases and large documents, and also to intelligently backtrack and re-evaluate its reasoning.” Opus 4.5 is anticipated to face significant competition from recently released advanced models, including OpenAI’s GPT 5.1 (released November 12th) and Google’s Gemini 3 (released November 18th).