Cost and usage issues on macOS — page 3
Issues the maintainers labelled both platform:macos and area:cost.
1,210 issues · 218 open · 992 resolved (82%) · first seen May 11, 2025
Is this getting better or worse?
This class of problem is converging. 258 new reports in the last 90 days vs 525 in the 90 before — -51%. The open backlog peaked at 275 in 2026-04 and sits at 218 today.
Derived by counting each matching issue from its creation date until it closed. First-party
computation over the anthropics/claude-code tracker — the trend is not published
anywhere upstream.
Which builds report this
843 of these reports name the Claude Code build they were running, spanning 192 releases. Heaviest reporters:
- Claude Code v2.1.119
- Claude Code v2.1.92
- Claude Code v2.1.198
- Claude Code v2.1.114
- Claude Code v2.1.123
- Claude Code v2.1.118
- Claude Code v2.1.201
- Claude Code v2.1.170
Report volume, not a defect count: an issue can mention a build without that build having caused the failure. Versions come from the version string in the bug-report template, matched against real published tags.
Do these get fixed?
Median time to close is 38 days across 992 closures. Of the 992 closures with a recorded reason, 16% were closed as completed and 831 as not-planned or duplicate. The most recent completed fix landed in 2026-10.
Workarounds reported by the community
98 of these issues post a workaround someone says worked. The highest-engagement ones:
- [BUG] Instantly hitting usage limits with Max subscription
- …communication that is inflating token usage significantly. Temporary fix for those on npm: npm install -g @anthropic-ai/claude-code@2.0.61 Hope this helps the devs pinpoint the regression.
Found in the thread of #16157 · still open - Need for dedicated .claudeignore file - .gitignore is insufficient for multi-platform projects
- …xclusion needs are context-dependent, not permanent Current Workaround Limitations The suggested solution of ["adding large build directories to your .gitignore file"]( doesn't work because: - These aren't build director…
Found in the description of #1304 · resolved - [BUG] API Error: Usage credits required for 1M context
- …session. Updating doesn't help (already on latest version). Workaround /model → standard context lets work continue, but 1M (included in Max) remains unusable.
Found in the thread of #63060 · still open - [BUG] 5-hour limit reached in less than 1h30
- ….3% , meaning ~20x token inflation per turn. After applying workarounds it stabilized at 89-99% . Full analysis with per-request measured data, safe workarounds, and community references (including [cc-cache-fix](
Found in the thread of #6457 · still open - [BUG] Opus usage limit reached unusually fast with Max 20x plan since Claude Code v2
- I found a workaround by manually setting Opus in plan mode in the settings, which seem to resolved the issue.(opusplan) Well.. i guess i have to contact support now...
Found in the thread of #8449 · resolved
Excerpts are quoted from the linked issue so you can judge relevance before opening it. Threads that only ask for a workaround, or report that there is none, are excluded. These are community suggestions, not official guidance from Anthropic — read the full thread before applying one.
Matching issues 201–300
Ranked by community engagement (reactions weighted over comments). Listing the top 300 of 1,210 matches; the statistics above cover all 1,210.
- [BUG] /usage weekly usage bar missing — only daily bar renders
- [BUG] [AGENT_TEAMS] Parallel Agent dispatch: 90min stall silently burns ~15M cache_read tokens, resets subagent context
- [Bug] Excessive context consumption on initial Opus startup (5% of pro plan limit)
- [BUG] Paid usage on max20 - limit reached within one command... WTF?!
- Usage Limit Reset Inconsistency with Documented 5-Hour Interval
- Unreadable Text Color in Light Theme for /cost and /model Commands
- [BUG] Irrecoverable Sessions if Toolcall canceled before completion
- You've hit your session limit · resets 12:30pm (America/Toronto)
- [BUGdeep-research workflow spawned 111 agents for a single query and consumed entire 5-hour session quota — zero output produced
- [Bug] Anthropic API Error: Usage credits required for 1M context window
- [BUG] Max plan: 'Usage credits required for long context requests' on standard 400k context
- [BUG] 20x Plan managed to burn the ENTIRE 5h window in ~30 minutes without any heavy use
- [Bug] Claude Code becomes unusable after reaching design token limits
- [Bug] ultrareview timeout shouldn't consume quota credits
- [BUG] double billed for claude max plan 2 months in a row and then account degraded to FREE PLAN
- [Bug] Excessive token consumption in long-running sessions
- You're out of extra usage · resets 4pm (America/New_York)
- [Bug] Excessive token consumption and slow performance
- [BUG] Different usage numbers in CLI/Cursor extension and Desktop app /Web
- [BUG] Model attempts file edits without reading file first, wasting tokens and time
- [Bug] Subscription Limit Tracking Calculation Error
- [BUG] Claude Code extension in VS Code is using more tokens than the CLI
- [BUG] Claude Code seems to use Sonnet even when configured explicitly to use Opus
- Feature request: a supported way to disable the agent_summary spinner (22.7% of requests, ~18% of weighted tokens in a subagent-heavy session)
- [Bug] Anthropic API Error: Usage credits required despite extended plan validity
- [BUG] Contradictory 'session limit resets at X' times across one session; background subagents die terminally and reset projections silently roll forward
- [BUG] Claude Max weekly usage depleted unexpectedly fast for Claude Code subagent-heavy Gmail MCP sessions
- [BUG] Urgent: Abnormal Token Consumption and Session Issue in Claude Code
- Workflow fan-out inherits a premium-tier default with no per-agent cost ceiling, causing ~$1k in auto-purchased charges
- [Bug] API usage calculation error: Single message consuming disproportionate quota
- Ultrareview: free credit consumed despite rate-limit failure (no report produced)
- [BUG] Usage limit triggers after a single prompt / continue request
- [BUG] 🚨 Excessive Token Usage in Claude, Small Prompts Consuming more than 80% Quota, Think before claude subscription
- [Bug] Prompt cache invalidation causing excessive token consumption on minor edits
- [BUG] Claude Code shows “Usage limit reached” even though my Usage page shows plenty of allowance remaining.
- [Bug] Unexpected quota consumption spike without usage pattern changes
- [BUG] 5h Limit on claude code cli at 100% and on claude code desktop on 3%
- Subagent cache miss on first SendMessage resume (cache_read=0)
- [BUG] Used all tokens in session limit without receiving a SINGLE token as a response
- [Bug] Excessive API token usage consumption
- [Bug] Token usage optimization for planning/non-code operations
- Regression ~2PM CST Dec 5: Aggressive credit consumption during background task polling
- [BUG] Claude Code extension gets permanently stuck on "clauding..."
- [Bug] Incorrect Model Usage for Default Sonnet Setting
- 5-hour limit reached ∙ resets 8pm
- [BUG] Token Usage Statistics Duplicated in stream-json Mode Causing Massive Cost Inflation
- [BUG] 5 hour limit reached after only 3 hours
- Claude Code repeats planning steps after the plan approval when running ![shell command] when planning is on and then asking for plan
- You've hit your session limit · resets 7pm (America/Toronto)
- Disproportionate subscription usage burn in long sessions: cache TTL expiry re-writes + no usage attribution in /usage
- [Bug] Excessive token consumption in session - 42% usage in 40 minutes for standard PR review workflow
- [BUG] Claude Code unusable on Pro plan due to 1M context bug — request for fix or credit
- [BUG] 1M context required error on active session — Max plan, Sonnet 4.6 High Effort, v2.1.157
- [Bug] Excessive token consumption in sessions
- [BUG] Excessive token consumption — task halted mid-execution with zero output (macOS)
- Opus 4.6 cache_read tokens counted at full rate despite cache hits on Max 20x
- [BUG] claude -p rejected with "out of extra usage" while Max plan 5h limit shows 35% remaining (not the HERMES.md trigger)
- [BUG] 5-hour usage limit exhausted after ~40 minutes of light use on Max 20x
- Claude Code Desktop silently uses project API keys for billing instead of subscription
- [Bug] Unexpected rapid token consumption with 5x plan
- [BUG] Incorrect Usage limit is reached popup even though it shows available
- Orphaned processes silently burn usage quota; dropped sessions continue running; unauthorized debit card charges
- [Bug] Excessive token consumption on simple queries (25k tokens for "what is 1+1")
- You've hit your limit · resets 5am (Europe/Istanbul)
- [Feature Request] Add transparent session limit tracking and advance warnings before hitting limits
- /usage dialog immediately dismisses without showing content on Max plan
- [BUG] Session usage counters reset unexpectedly mid-cycle
- [Bug] Phantom shell process counts against 5-hour daily limit after KillShell confirmation
- [BUG] Input token reported as 0 in `/cost`
- [Bug] Anthropic API: Unexpected Usage Depletion on MAX20 Plan
- Token Usage Regression in /context Command Between v1.0.86 and v1.0.88
- Claude AI usage limit reached|1749855600
- [BUG] Cost increased a lot
- [Bug] Claude Fable 5 incorrectly consumes usage credits on Claude Code 2.1.215 with 20x Max subscription
- [BUG] Fable 5 requires API credits before metering on Jul 19
- [Bug] Uncontrolled recursive agent spawning causes excessive API usage and costs
- [Bug] Excessive cache writes causing unexpectedly high token usage and costs
- [BUG] within 1 hour weekly usage consumed 40 percent, it only used rest 45 due to rough coding in 2 days , some issue
- Runaway output: model repeats fixed string "court placeholder" thousands of times, exhausting output tokens (long-running session, large context)
- [BUG] ScheduleWakeup persists after Ctrl+C, daemon auto-respawns loop unattended causing unbounded token spend
- Unprocessable image in context burns credits with no recovery path
- [Bug] Excessive token consumption from repeated echo tool calls
- [BUG] Workflow harness retries indefinitely on HTTP 429 rate_limit — 97 agents, 2M tokens burned in 34 seconds ()
- [BUG] Dispatch conductor session grows unbounded with no rotation/compaction, forcing premium 1M-context billing
- [BUG] agent thinking work lost when hitting limit
- [BUG]Advisor tool injects same-tier opus calls into every opus subagent, doubling spend invisibly
- [BUG] /usage no longer renders visual horizontal bar graphs in TUI
- [BUG] Usage Calculation for weekly limits hit within a short time with no large use.
- Massive token burn: 80% → 11% in ~5 minutes on a single session
- Saying I hit my limit when I haven't (just upgraded plan)
- [BUG] [ultrareview] Remote review returns empty findings ([]) on large-scope repositories
- You've hit your limit · resets 1am (Asia/Makassar)
- [BUG] Increase in usage calculations and cannot submit using /bug
- [BUG] 2x - 3x increase in output tokens starting with version 2.1.96 and increasing in 2.1.101
- [BUG] Hidden `isMeta` system-reminder causes 200k+ cache_creation burst mid-session (Opus 4.6)
- [BUG] Quality of Claude Code is atrocious. I've paid hundreds for this service
- Model behavior degradation: factual fabrication, context loss, document access failures — claude-sonnet-4-6, Cowork v1.2581.0 (Mac, April 2026)
- [FEATURE] Token usage visibility and threshold hooks in Claude Code
- You've hit your limit · resets 5pm (Europe/Brussels)
- [BUG] 5-minute session exhaustion on Max plan using v2.1.87 with Claude in Chrome