Claude Opus 4.7 Is In The Picker

April 16, 2026

Anthropic's flagship model is now available in Alfrada — frontier reasoning, a 1M-token context window, and the ability to think harder when the question warrants it instead of burning the same budget on every prompt. Pro Max and Black tiers can pick it now.

What's new

• Claude Opus 4.7 appears in the model selector, ready to pick for any session.
• 1M-token context window — fits a full codebase, a long research corpus, or hours of meeting transcript with room left.
• Adaptive thinking — the model decides how deeply to reason. A simple question gets a fast answer; a hard one gets a long, careful one. No fixed thinking budget.
• Swarm-capable — useful as the brain of a Swarm that delegates to cheaper worker models.
• Opus 4.6 and Sonnet 4.6 also got adaptive thinking — same reasoning-scaling behaviour at a lower price point.

When you'd reach for it

• A large codebase refactor where the model needs to keep multi-file dependencies straight while planning a migration.
• A complex investment memo where you want deep thinking on the hard parts (market structure, competitor response) and quick answers on the factual ones.
• A multi-day research project run as a Swarm — Opus 4.7 leads, cheaper models do the legwork.

Try it

• "Plan a migration for this codebase and walk me through every multi-file dependency."
• "Write an investment memo on this target — go deep on competitive dynamics."
• "Run a research swarm: leader plans and synthesizes, workers scrape and draft."

Heads up

Opus 4.7 is the priciest model the auto-router will pick. If you want to cap spend on routine work, the workhorse-tier models will handle most of it at roughly a fifth of the cost — manually pick one or let auto-routing keep you out of frontier-tier when the prompt doesn't need it. Available on Pro Max, Black, Pro Max VIP, and Black VIP.