Right on cue with the Opus 5 launch, Claude Code 2.1.219 shipped — and this one is more than a routine update. It packs 27 CLI changes, but three of them stand out.
Opus 5 becomes the new default
The most important item hides in a single changelog line: the opus alias and the default now resolve to Opus 5 — on Max, Team Premium, Enterprise pay-as-you-go and via the API. If you’ve simply been using ‘Opus,’ you’re on the new model after the update without doing a thing. There’s also a fast mode with a 1M context at $10 input and $50 output per million tokens.
That’s convenient — but also worth watching. The model changed; your settings didn’t. If you’re comparing results or depend on a specific behavior, it’s worth a quick check of which model is actually running.
Bash tool and subagents
Two new capabilities lift this release above cosmetic tweaks. First, a tool that executes bash commands and returns their output directly — so Claude can pull shell data on the spot instead of collecting it the hard way. Second, the ability to launch agents for complex, multi-step tasks, letting you delegate and compose subtasks — a step closer to real orchestration.
The rest of the 27 changes is the usual mix of fixes and polish you’ve come to expect from Claude Code on an almost daily basis.
My take
The interesting part is the combination: a cheaper, stronger model becomes the default just as the CLI gains tools to spread work across subagents. Together, those two make longer-running, nested tasks suddenly affordable and practical. Only the silent model switch stays double-edged — convenience on one side, less control on the other. My tip: after updating, deliberately check which model you’re actually running.
Sources: