claude-code - 💡(How to fix) Fix [Bug] Claude Code: Model generating inconsistent responses and consuming excessive token quota on single prompts [1 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#52722Fetched 2026-04-24 10:41:24
View on GitHub
Comments
0
Participants
1
Timeline
5
Reactions
0
Author
Participants
Timeline (top)
labeled ×5

Code Example

[]
RAW_BUFFERClick to expand / collapse

Bug Description Lying, contradicting own responses, misinterpreting my questions, assuming things, making up numbers and taking a very long time to think. Ate up my ~40% session quota in single prompt. Overall extremely frustrating session. Each code run takes 6-7 hours, should I trust the model with all the code changes or lose so much time? I am not sure anymore. I am starting to have doubts now. I will not use Claude for sometime now.

Environment Info

  • Platform: darwin
  • Terminal: Apple_Terminal
  • Version: 2.1.109
  • Feedback ID: 7fbced76-15d3-422c-a4c7-94872a257cea

Errors

[]

extent analysis

TL;DR

Re-evaluate the trust in the model's code changes and consider alternative approaches to minimize time loss.

Guidance

  • Assess the model's performance and accuracy in previous code runs to determine if the 6-7 hour runtime is justified.
  • Consider breaking down large code changes into smaller, more manageable chunks to reduce the time spent on each run.
  • Evaluate the trade-off between trusting the model with all code changes and manually reviewing each change to ensure accuracy.
  • Look into optimizing the development workflow to reduce the overall session quota usage.

Notes

The issue lacks specific technical details about the code changes and the model's behavior, making it difficult to provide a more targeted solution.

Recommendation

Apply workaround: given the frustration and time consumption, it's reasonable to explore alternative approaches to using the model for code changes, at least until the issues are better understood or resolved.

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…

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 Code: Model generating inconsistent responses and consuming excessive token quota on single prompts [1 participants]