What is Opus?

Opus is Anthropic’s most powerful AI model. It’s the top tier in their model lineup, which goes Haiku (fastest, cheapest), Sonnet (balanced), and Opus (most capable). If you need an AI to handle really complex reasoning, long documents, or difficult coding tasks, Opus is what you’d reach for.

Anthropic names their models after literary forms. Haiku is short and fast. Sonnet is more substantial. Opus is the most elaborate and powerful. Each tier has different versions as the models get updated. As of February 2026, the latest is Opus 4.6, which came out on February 5, 2026.

Opus models cost significantly more than Sonnet or Haiku because they use more compute. Opus 4.6 pricing is $5 per million input tokens and $25 per million output tokens. For context, that makes it one of the most expensive LLMs on the market, but also one of the most capable.

What makes Opus stand out is performance on hard tasks. On Anthropic’s internal performance engineering exam, Opus 4.5 scored higher than any human candidate had ever scored. Opus 4.6 leads the industry on agentic coding, tool use, and complex enterprise workflows. It also supports a million-token context window in beta, which means it can work with massive amounts of text at once.

The tradeoff is speed and cost. For most everyday tasks, Sonnet is usually the better choice. But when you need the absolute best reasoning and you’re willing to pay for it, Opus is the model to use.

I’ve been using Opus all day through Claude Code for work and for building Code Q&A. It’s miraculous. The difference in capability compared to Sonnet is noticeable when you’re doing complex coding tasks.