Issue Station
Sign In
Tools hereToolbox

claude-code - 💡(How to fix) Fix [BUG] Claude Opus 4.7 thinking tokens redacted even with showThinkingSummaries or similar flags set to true [3 comments, 3 participants]

Official PRs (…)
ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

Helpful · Quick feedback

Loading…
GitHub stats
anthropics/claude-code#49622•Fetched 2026-04-17 08:35:58
View on GitHub
Comments
3
Participants
3
Timeline
8
Reactions
1
Author
BraxbroBraxbro
Participants
BraxbroBraxbrogithub-actions[bot]github-actions[bot]solatissolatis
Timeline (top)
labeled ×4commented ×3subscribed ×1

Error Message

Error Messages/Logs

RAW_BUFFERClick to expand / collapse

Preflight Checklist

  • I have searched existing issues and this hasn't been reported yet
  • This is a single bug report (please file separate reports for different bugs)
  • I am using the latest version of Claude Code

What's Wrong?

Since the release of Opus 4.7, all thinking tokens of 4.7 are redacted entirely - thinking blocks are replaced with blank strings with token count receipts and the like rather than containing any useful info. I use the latest version of Claude Code for VS Code.

This is an extremely damaging breaking change to my workflow, which depends on real-time auditing of thinking to avoid wasted tokens on unproductive looping thinking, to track the flow of thought with relation to tool calls and file reads to carefully manage context poisoning, and allows me to stop the model and interject with useful missing context when it's clearly lacking something.

This is critical to both managing costs and is one of the main differentiators that keeps me returning to the service.

What Should Happen?

When responding, Opus 4.7 should offer the thinking dropdown as normal, and include summaries (if not full output) of the model's thought process to enable the above usecases. While I realize the measure is likely taken to prevent model distillation, it also massively hampers auditability of model operations.

Alternatively, Opus 4.6 should be re-added to the model picker to allow users dependent on such functionality to still use Opus with full VS Code integration without losing this feature.

Error Messages/Logs

N/A - the absence of any output is the problem

Steps to Reproduce

Open Claude Code in VS Code, set your model to Opus 4.7, and send a message. Even with verbose output or similar settings enabled, the model responds with empty thinking token blocks.

Claude Model

Opus

Is this a regression?

Yes, this worked in a previous version

Last Working Version

Not sure - it still functions for Haiku and Sonnet, and was functional for 4.6 as of April 14.

Claude Code Version

2.1.112

Platform

Anthropic API

Operating System

Ubuntu/Debian Linux

Terminal/Shell

VS Code integrated terminal

Additional Information

N/A

extent analysis

TL;DR

The issue can be potentially resolved by re-adding Opus 4.6 to the model picker or finding an alternative way to access the thinking token output in Opus 4.7.

Guidance

  • Verify that the issue is specific to Opus 4.7 by testing other models like Haiku and Sonnet, which are reported to still function correctly.
  • Check the Claude Code settings in VS Code to ensure that verbose output or similar settings are enabled, which might help in retrieving the thinking token blocks.
  • Consider reaching out to the developers or support team to request the re-addition of Opus 4.6 to the model picker as a temporary workaround.
  • Investigate if there are any configuration options or flags in the Anthropic API that could allow access to the thinking token output in Opus 4.7.

Example

No code snippet is provided as the issue seems to be related to the model's behavior rather than a specific code implementation.

Notes

The issue seems to be a regression introduced in Opus 4.7, and the exact cause is not clear. The solution might involve changes to the model or the Claude Code integration.

Recommendation

Apply workaround: Request the re-addition of Opus 4.6 to the model picker or explore alternative models that provide the required functionality, as this seems to be the most straightforward way to restore the previous behavior.

Vote matrix · Quick signals

Works
Did the solution work? Tap to confirm.
Easy Fix
Was it a quick fix?
Time Saver
Did it save you time?
Blocking
Was it severely blocking?
Common Issue
Are others likely hitting this too?
Flaky / Intermittent
Is it intermittent?
Verified / Reproducible
Can you reproduce it reliably?
Loading…
#api#docker error#permission error#memory optimization#batch processing

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Back to top recommendations

TRENDING

claude-code - 💡(How to fix) Fix [BUG] Claude Opus 4.7 thinking tokens redacted even with showThinkingSummaries or similar flags set to true [3 comments, 3 participants]