Opus 4.7 is now the default on Cyrus

Anthropic recently released Claude Opus 4.7, the successor to Opus 4.6. Starting today, Opus 4.7 is the default model on every Cyrus agent.

What's new in Opus 4.7

Opus 4.7 builds directly on the foundation Opus 4.6 laid down and pushes further on the things that matter most for agentic coding: judgment, reliability, and sustained focus across long tasks. Where 4.6 redefined what a coding model could do, 4.7 sharpens the blade.

The improvements we've noticed most:

• Stronger agentic reliability: better moment-to-moment decisions inside long tasks — when to keep pushing, when to pause and verify, and when to ask for clarification instead of guessing
• Better code review and self-debugging: 4.7 is noticeably sharper at catching its own mistakes, spotting subtle bugs, and converging on correct solutions in fewer iterations
• More consistent planning: stays on track across multi-step work without drifting, even in large, unfamiliar codebases
• Sharper judgment on tradeoffs: more thoughtful about architecture decisions and when a "quick fix" is actually a bad fix that will cost you later

Put simply: fewer iterations, fewer rewrites, and fewer PRs that need a human to untangle them before they can land.

What this means for your team

For product teams using Cyrus to automate development workflows, Opus 4.7 translates directly into velocity. Issues that used to need a second or third round now land correctly on the first pass. PRs come back cleaner. Edge cases that earlier models missed get caught during the agent's own review, before a human ever looks at the diff.

Combined with Cyrus's swarm mode and Linear-native workflow, Opus 4.7 means more of your backlog clears itself overnight — and the work that comes back is the kind you'd actually ship, not drafts that need rewriting.

If your team has been bottlenecked on review capacity, this is the upgrade that matters most. The agent's output quality is what determines whether autonomous work is a force multiplier or just more noise in your queue. Opus 4.7 raises that bar.

Getting started

For cloud-hosted Cyrus users, no action required. Every Cyrus agent is now running on Opus 4.7 by default. Your existing workflows, Linear integrations, and agent configurations continue to work exactly as before — just with a more capable model behind them.

For self-hosted Cyrus users, update to the latest release to pick up Opus 4.7 as the default model.

If you're not yet using Cyrus, get started today and see what the strongest coding model available can do for your team.