Claude Opus 4.6: 1M context, agent teams, and leadership on Terminal-Bench 2.0
In one sentence Anthropic releases Opus 4.6: first Opus with 1M-token context in beta, agent teams in Claude Code (research preview), stated top scores on Terminal-Bench 2.0 and Humanity's Last Exam. Pricing unchanged at $5/$25.
Opus is Anthropic's "premium" model, the one to use when a task is genuinely hard. Version 4.6 lands on February 5, 2026 and brings three main changes.
First: context goes up to 1 million tokens in beta, for the first time on an Opus-class model. That means passing entire document archives or large codebases without trimming.
Second: "agent teams" in Claude Code, as a research preview. Instead of a single agent working sequentially, several agents work in parallel and coordinate on their own; Anthropic says they work best on independent, read-heavy tasks such as codebase reviews.
Third: according to Anthropic it is the best model on Terminal-Bench 2.0 (agentic coding) and Humanity's Last Exam (multidisciplinary reasoning). Pricing unchanged: $5 input, $25 output per million tokens.
Companies
Anthropic
Tools
Claude Opus 4.6, Claude Code, Claude in Excel, Claude in PowerPoint
Tags
Sources