Anthropic launches Claude Opus 4.6
What's the story
Anthropic has unveiled its most advanced AI model, Claude Opus 4.6, with a groundbreaking one million token context window. This new capability allows the AI to process and retain massive amounts of information in one go, making it ideal for long-running, autonomous tasks across large codebases. The model also introduces 'agent teams,' enabling multiple instances of Claude to collaborate on complex tasks simultaneously.
Performance boost
Opus 4.6 outperforms GPT-5.2 in various domains
Claude Opus 4.6 has been specifically designed for software development, financial analysis, research, and document creation. It can catch its own errors and sustain long engineering sessions, areas where AI reliability often falters. The model has already outperformed OpenAI's GPT-5.2 in finance, legal, and research domains and topped Terminal-Bench 2.0 coding evaluation with its superior capabilities in coding/debugging/software development tasks.
Feature highlight
'Agent teams' for large-scale AI collaboration
The 'agent teams' capability is the standout addition in Claude Opus 4.6. It allows multiple instances of the model to work together on complex tasks, enabling large-scale AI collaboration. This feature builds on Claude Cowork, Anthropic's environment for autonomous task execution, where Opus 4.6 can perform extended work with minimal human supervision.
Safety measures
Improvements in safety and cybersecurity assessments
On the safety front, Anthropic claims that Opus 4.6 shows fewer unnecessary refusals and reduced misaligned behavior like deception or hallucination. It also introduces new cybersecurity assessments that evaluate both defensive and offensive capabilities, ensuring the model operates within safe and predictable limits. These improvements reflect Anthropic's commitment to artificial general intelligence (AGI), a future where machines could handle virtually all forms of intellectual work.
Twitter Post
1st Opus-class model with one million token context in beta
Introducing Claude Opus 4.6. Our smartest model got an upgrade.
— Claude (@claudeai) February 5, 2026
Opus 4.6 plans more carefully, sustains agentic tasks for longer, operates reliably in massive codebases, and catches its own mistakes.
It’s also our first Opus-class model with 1M token context in beta. pic.twitter.com/L1iQyRgT9x