Anthropic's flagship model — strong agentic coding, long-horizon reasoning and careful instruction following.
Specifications
- Provider
- Anthropic
- Type
- Vendor / proprietary
- Modality
- Text + Vision
- Category
- Reasoning model
- Context window
- 1M
- Knowledge cutoff
- Early 2026
- Released
- May 27, 2026
What it was trained for
Anthropic's flagship Opus-tier model, built for highly autonomous, long-horizon work with strong agentic execution, coding, and knowledge tasks.
Best for
- ▸Agentic coding and large multi-file refactors
- ▸Long-horizon autonomous agent loops
- ▸Complex reasoning and analysis
- ▸Knowledge work and document editing
- ▸Code review and debugging
- ▸Tool-using and multi-agent systems
Capabilities
Vision inputFunction calling / toolsAdaptive thinkingAdjustable effort levels1M contextStructured outputs
Performance & positioning
Anthropic's most capable model to date, excelling at long-horizon agentic execution, coding, and memory.
More from Anthropic