Anthropic launches Claude Opus 4.5 to take on ChatGPT and Gemini

In a bold move following the launches of GPT-5.1 and Gemini 3, Anthropic has introduced its latest AI model, Claude Opus 4.5. The startup asserts that this new model is the leading solution globally for coding, agent tasks, and various computer-related functions. Claude Opus 4.5 has made a significant mark by achieving an impressive score of 80.9% on the SWE-bench Verified, a benchmark that evaluates real-world software engineering capabilities. This score not only sets a new standard as the first model to surpass the 80% threshold but also positions it ahead of competitors; Google’s Gemini 3 Pro scored 76.2%, while OpenAI's GPT-5.1 Codex Max achieved 77.9%. Furthermore, the model has demonstrated its superiority over human candidates in a rigorous two-hour assessment designed for performance engineering applicants. According to Anthropic, this test evaluates technical acumen and decision-making under pressure, indicating a significant shift in how AI could reshape the engineering profession. In addition to its coding prowess, Claude Opus 4.5 excels in the τ2-bench, a benchmark focusing on agent performance in complex real-world scenarios. One notable test involved acting as an airline service agent responding to a distressed customer, where the model successfully navigated booking restrictions through an innovative approach, showcasing its problem-solving capabilities. Anthropic also emphasizes that Claude Opus 4.5 is their most aligned model to date, boasting enhanced defenses against prompt injection attacks, which can manipulate AI responses. The company claims that the new model is more resistant to such tactics than any other leading AI on the market. Available on the Claude app for both Android and iOS users, as well as on the Claude website, this new model is also being rolled out to developers, expanding its reach in the AI landscape.