Why did Claude Code start using Opus instead of Sonnet for me?

Claude Code 2.1.280 changed the default model on Pro and Team Standard plans from Sonnet to Opus, which brings those plans in line with Max, Team Premium and Enterprise. It is a default, not a lock — /model still switches the session to another model, and the choice persists the way it always has.

How much does Claude Opus 5.5 cost in Claude Code?

Anthropic lists Claude Opus 5.5 at $4 per million input tokens and $20 per million output tokens, with cache reads at $0.20 per million. The model carries a 1M-token context window and is registered in Claude Code as claude-opus-5-5, where 2.1.280 makes it the default Opus model.

What was the symlink write problem this release fixes?

Claude Code used to judge a write by the path as spelled inside the project, so a path that passed through a symlink could be approved even though the file landed somewhere else. In 2.1.280 the permission prompt names where the write actually lands, and acceptEdits, allow rules and auto mode no longer approve one that resolves outside the project.

Why is my saved effort level not being used on Opus 5.5?

Claude Code 2.1.280 stops an effort level saved before /effort became per-model from carrying over to newly released models. Opus 5.5 starts at its own default level until a level is picked for it. In the same change, Opus 4.7, Opus 4.8 and Fable 5 no longer hold their launch-default effort over a level set through /effort in -p, the Agent SDK, or an effortLevel setting.

Did anything change for auto mode in this release?

Yes, two loops were stopped. When a safety check declines to review an action, Claude Code 2.1.280 denies it once and says retrying will not help, instead of retrying repeatedly. When a check returns no answer, retries back off and the turn stops with a message after ten in a row rather than denying without pause.

Anthropic · 2026-09-22 · major

Claude Code 2.1.280 — Opus 5.5 arrives and Pro plans move off Sonnet

Claude Code 2.1.280 adds Claude Opus 5.5 as the default Opus model and changes the default on Pro and Team Standard plans from Sonnet to Opus. A permission fix stops writes through a symlink landing outside the project unchecked.

Claude Opus 5.5 becomes the default Opus model, and Pro and Team Standard plans now start on Opus instead of Sonnet.

Quick facts

Version	v2.1.280
Maker	Anthropic
Released	2026-09-22
New model	Claude Opus 5.5 (claude-opus-5-5), now the default Opus
Opus 5.5 price	$4 / $20 per MTok, $0.20 cache reads
Opus 5.5 context	1M tokens
Default model change	Pro and Team Standard move from Sonnet to Opus

What is it?

Claude Opus 5.5 lands in Claude Code 2.1.280 as the default Opus model, with a 1M-token context window at $4 and $20 per million tokens and $0.20 per million for cache reads. The same release changes which model a session starts on: Pro and Team Standard plans now default to Opus rather than Sonnet, matching what Max, Team Premium and Enterprise already did.

How does it work?

Effort settings were untangled to go with the new model. An effort level saved before /effort became per-model no longer applies to newly released models such as Opus 5.5 — they start at their own default until a level is picked — and Opus 4.7, Opus 4.8 and Fable 5 stop holding their launch-default effort over a level set through /effort in -p, the Agent SDK, or a project, managed or --settings effortLevel. Auto mode also stops spinning: an action a safety check declined to review is now denied once with a note that retrying will not help, and unanswered checks back off before the turn stops after ten in a row.

Why does it matter?

The permission fix is the one to read twice. Writes through a symlinked path used to be judged by the path as typed rather than where the file actually lands, so acceptEdits, allow rules and auto mode could approve a write that ended up outside the project; the prompt now names the real destination and those rules no longer approve it. Elsewhere, CLAUDE_CODE_MAX_MCP_DESCRIPTION_LENGTH lifts the 2,048-character cap on MCP tool descriptions for a session, and the VS Code extension gains typed /status, /sandbox, /chrome, /export, /skills and /plan dialogs.

Who is it for?

Claude Code users on Pro, Team and Enterprise plans

Frequently asked questions

Why did Claude Code start using Opus instead of Sonnet for me?: Claude Code 2.1.280 changed the default model on Pro and Team Standard plans from Sonnet to Opus, which brings those plans in line with Max, Team Premium and Enterprise. It is a default, not a lock — /model still switches the session to another model, and the choice persists the way it always has.
How much does Claude Opus 5.5 cost in Claude Code?: Anthropic lists Claude Opus 5.5 at $4 per million input tokens and $20 per million output tokens, with cache reads at $0.20 per million. The model carries a 1M-token context window and is registered in Claude Code as claude-opus-5-5, where 2.1.280 makes it the default Opus model.
What was the symlink write problem this release fixes?: Claude Code used to judge a write by the path as spelled inside the project, so a path that passed through a symlink could be approved even though the file landed somewhere else. In 2.1.280 the permission prompt names where the write actually lands, and acceptEdits, allow rules and auto mode no longer approve one that resolves outside the project.
Why is my saved effort level not being used on Opus 5.5?: Claude Code 2.1.280 stops an effort level saved before /effort became per-model from carrying over to newly released models. Opus 5.5 starts at its own default level until a level is picked for it. In the same change, Opus 4.7, Opus 4.8 and Fable 5 no longer hold their launch-default effort over a level set through /effort in -p, the Agent SDK, or an effortLevel setting.
Did anything change for auto mode in this release?: Yes, two loops were stopped. When a safety check declines to review an action, Claude Code 2.1.280 denies it once and says retrying will not help, instead of retrying repeatedly. When a check returns no answer, retries back off and the turn stops with a message after ten in a row rather than denying without pause.

Try it

npm i -g @anthropic-ai/[email protected]