claude-code - 💡(How to fix) Fix Claude Max subscribers are not getting what they are paying for — thinking budget delivery, effort persistence, and communication are all broken [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#46838Fetched 2026-04-12 13:31:44
View on GitHub
Comments
0
Participants
1
Timeline
3
Reactions
1
Participants
Timeline (top)
labeled ×3
RAW_BUFFERClick to expand / collapse

I am a Claude Max subscriber. I build 12-16 hours a day on this product. I pay for it expecting to use the best version of it.

Here is what I have found:

1. /effort max cannot be persisted. Every new session resets to high. Running /effort max explicitly tells you "this session only." There is no documented way to make max effort the permanent default in settings.json. The only persistent value appears to be high. If I am paying for Claude Max, why is the highest effort level intentionally locked to session-scope only?

2. "Max" may not mean anything. GitHub issue #20350 documents — via independent MITM traffic analysis — that Claude Code delivers approximately 10% of the requested thinking budget regardless of what effort level is set or what subscription tier the user is on. If that is accurate, the effort levels are theater. I am paying for max and getting 10% of it.

3. The communication is not there. I understand Anthropic made a deliberate decision here. I understand the compute economics. That is not my complaint. My complaint is that I found out about this by accident. There was no notice to Claude Max subscribers that effort persistence was session-scoped, no explanation of what thinking budget I am actually receiving, and no transparency about what "max" actually delivers.

The ask:

  • Make effortLevel: "max" a valid persistent value in settings.json
  • Publish what thinking token budget each effort level actually delivers per subscription tier
  • When defaults or delivery behavior change, communicate it directly to affected users — especially Claude Max

extent analysis

TL;DR

To address the issue, the developer should consider making effortLevel: "max" a persistent value in settings.json and provide transparency on the thinking token budget for each effort level and subscription tier.

Guidance

  • Review the settings.json documentation to confirm if there are any plans to add effortLevel: "max" as a valid persistent value.
  • Investigate the claims made in GitHub issue #20350 regarding the actual thinking budget delivered by Claude Code, regardless of effort level or subscription tier.
  • Consider adding a feature to communicate changes in defaults or delivery behavior directly to affected users, especially Claude Max subscribers.
  • Evaluate the compute economics and deliberate decisions made by Anthropic to understand the reasoning behind the current implementation.

Example

No code snippet is provided as the issue does not contain specific code-related details.

Notes

The issue highlights the need for transparency and communication regarding the actual benefits of the "max" effort level and the thinking token budget for each subscription tier. The developer should prioritize addressing these concerns to improve the user experience for Claude Max subscribers.

Recommendation

Apply a workaround by manually setting the effort level to "max" at the start of each session until a permanent solution is implemented, as the current implementation does not support persistent "max" effort levels. This is because the issue is more related to the lack of transparency and communication rather than a technical bug that can be fixed with an upgrade.

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