Anthropic releases Claude Opus 4.6 featuring a 1M token context window and advanced coding capabilities that discovered over 500 zero-day vulnerabilities during testing, setting new standards for AI security.