claude-code - 💡(How to fix) Fix Phantom ~22K tokens per session on one account vs another — same machine, same project [1 comments, 2 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#45515Fetched 2026-04-09 08:03:38
View on GitHub
Comments
1
Participants
2
Timeline
6
Reactions
0
Timeline (top)
labeled ×4commented ×1cross-referenced ×1
RAW_BUFFERClick to expand / collapse

Two Claude Code Max subscriptions on the same machine (WSL2 Linux), same project, same settings, same CC version. The only difference: which claude.ai account is logged in.

What happens

On every prompt (even simple 1+1), Login #1 (old account) consumes ~22K more tokens than Login #2 (new account). This is visible in the status bar (real API billing), not just /context.

The extra tokens do not appear in any /context category — they're invisible. All categories sum correctly on Login #2 but leave a ~22K gap on Login #1.

Login #1 (old)Login #2 (new)
Status bar (simple 1+1)higherlower
/context header~78.9K~56.2K
/context category sum~56.2K~56.2K
Phantom gap~22K unaccounted~0

Categories sum to ~56.2K on both logins — but Login #1's header reports ~78.9K. The ~22K difference is invisible in the breakdown.

Screenshots

/context side-by-side + simple prompt test:

(screenshot 1 — will be attached)

Settings comparison (identical on both) + math test:

(screenshot 2 — will be attached)

What we ruled out

HypothesisTestResult
claude.ai MCP connectors (Asana 47 tools, Gmail, Calendar, n8n)Disabled all on Login #1Gap stayed
claude.ai MemoryDisabled + cleared on bothNo change
Grove / "Help improve Claude"Checked — identical on bothNot the cause
CC local settingsDiffed — identicalNot the cause
JSONL user contentCompared — identical payloadNot the cause
Deferred tool schemasSame hash, same 196 tools, 71 skillsNot the cause

Environment

  • Claude Code v2.1.97, Linux (WSL2), same machine
  • 12 project MCP servers (identical config on both logins)
  • Login #1 additionally has 4 claude.ai connectors (but disabling them didn't close the gap)
  • Both accounts: Max subscription, claude.ai Memory OFF

Impact

~22K phantom tokens per session = ~2% of 1M context window consumed by something invisible and uncontrollable. On long sessions with many turns, the cumulative overhead grows. Users have no way to see what these tokens are or reduce them.

extent analysis

TL;DR

Investigate account-specific settings or claude.ai backend differences to address the ~22K token discrepancy between Login #1 and Login #2.

Guidance

  • Verify that all account settings, including those not explicitly mentioned, are identical for both logins to rule out any overlooked configuration differences.
  • Check with claude.ai support to see if there are any known issues or differences in how tokens are calculated or billed for different accounts, especially considering the "Max subscription" status.
  • Attempt to replicate the issue on a different machine or environment to determine if the problem is specific to the WSL2 Linux setup or if it's account-related.
  • Review the documentation for Claude Code and claude.ai to ensure that all features and settings are fully understood, particularly those related to token usage and billing.

Example

No specific code snippet can be provided without more details on the API or the exact implementation of the Claude Code and claude.ai integration. However, any debugging efforts should focus on comparing the API requests and responses for both logins to identify any differences that could explain the token discrepancy.

Notes

The issue seems to be related to the difference in how tokens are billed or calculated for the two logins, rather than a problem with the Claude Code or WSL2 setup itself. The fact that disabling certain features and connectors did not resolve the issue suggests that the root cause might be related to the accounts or the claude.ai backend.

Recommendation

Apply workaround: Since the exact cause is not clear and seems related to account-specific settings or backend differences, the best immediate action might be to use Login #2 (new account) for critical tasks to avoid the extra token consumption, while continuing to investigate the root cause with claude.ai support.

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