[BUG] Opus model lacks clear activity/completion indicators in CLI
Preflight Checklist
- [x] I have searched existing issues and this hasn't been reported yet
- [x] This is a single bug report (please file separate reports for different bugs)
- [x] I am using the latest version of Claude Code
What's Wrong?
When using Opus models in the CLI/TUI, status indicators (such as the spinner verbs and the final "thought for x sec" line) are not consistently shown; the completion indicator is especially inconsistent. This makes it extremely difficult to tell when Opus is actually done working on a task, or if it is still running, or if it is stuck.
This was also reported here but as specific to VS Code; this is happening for me in CLI as well.
What Should Happen?
Consistent status indicators are shown, including when Claude is still thinking and when Claude is done thinking/responding, so that the user knows when to wait for Claude and when they can respond to Claude.
Steps to Reproduce
- Start Claude Code in terminal and switch to Opus model
- Ask Claude to perform a task that will take some amount of effort (especially if it's something that uses sub-agents)
- See that status indicators appear and work inconsistently
- Switch back to Sonnet model
- Ask Claude to perform the same or a similar task
- See that status indicators now appear and work reliably
Claude Model
Opus
Claude Code Version
2.1.85
Platform
Anthropic API
Operating System
macOS
Terminal/Shell
Terminal.app (macOS)
Additional Information
I even asked Claude about it, hoping it was something I might be able to configure locally, and it agreed that this as a reasonable thing to want. 😄
Unfortunately, this isn't something I can configure from my side — it's a harness/UI behavior. It would be worth filing as feedback at https://github.com/anthropics/claude-code/issues if you'd like the team to look at making the activity indicators more consistent across models. The core ask — "I want a clear signal when the model is done" — is very reasonable.
This issue has 2 comments on GitHub. Read the full discussion on GitHub ↗