Cost and usage issues on Linux — page 2
Issues the maintainers labelled both platform:linux and area:cost.
277 issues · 53 open · 224 resolved (81%) · first seen May 8, 2025
Is this getting better or worse?
This class of problem is holding steady. 96 new reports in the last 90 days vs 113 in the 90 before — -15%. The open backlog peaked at 69 in 2026-04 and sits at 53 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
211 of these reports name the Claude Code build they were running, spanning 108 releases. Heaviest reporters:
- Claude Code v2.1.119
- Claude Code v2.1.207
- Claude Code v2.1.126
- Claude Code v2.1.87
- Claude Code v2.1.92
- Claude Code v2.1.150
- Claude Code v2.1.123
- Claude Code v2.1.237
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 37 days across 224 closures. Of the 224 closures with a recorded reason, 20% were closed as completed and 180 as not-planned or duplicate. The most recent completed fix landed in 2026-08.
Workarounds reported by the community
34 of these issues post a workaround someone says worked. The highest-engagement ones:
- [BUG] Conversation history invalidated on subsequent turns
- …put tokens, which can cause a massive spike in token usage. Workarounds to minimize cache invalidation: 1. Keep conversations shorter : Start new sessions for new tasks rather than continuing a single long session. Short…
Found in the thread of #40524 · resolved - CC v2.1.100+ inflates cache_creation by ~20K tokens vs v2.1.98 — same payload, server-side
- …r potentially degraded output, with no visibility into why. Workaround Downgrade to v2.1.98 or earlier. Check for available binaries, or use . Note: Auto-updates may overwrite older versions. v2.1.98 is no longer retaine…
Found in the description of #46917 · still open - [Bug] Excessive token usage in Claude Code 2.1.1 - 4x+ faster rate consumption than previous versions
- …-- but until that ships, parsing your own logs is the best workaround.
Found in the thread of #16856 · still open - [BUG] Prompt cache regression in --print --resume since v2.1.69(?): cache_read never grows, ~20x cost increase
- …advertently broken cache breakpoint placement for sessions. Workaround Pinned to v2.1.68 ( ). What Should Happen? Expected behavior (v2.1.68) grows as conversation accumulates, drops to a small delta (~800 tokens): Actua…
Found in the description of #34629 · resolved - Max Plan: Token usage increased ~3-5x without any configuration change (since ~March 28-29)
- ….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 #41506 · still open
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 101–200
Ranked by community engagement (reactions weighted over comments).
- [BUG] The conversation is summarized below: Claude AI usage limit reached
- [BUG] Image Inputs via Read Tool or @ Reference Consume 7–10× More Tokens with On-Prem LLMs
- [BUG] Weekly "all models" counter dropping mid-cycle without reset (Max, Opus 4.7)
- Reopening: "Continue" leads to massive context carry-over into new 5h limit, thus when implementation stops half-way and you are asked to wait, the next session quota is reduced massively.
- [Bug] Claude Code not recognizing plan upgrade, continues using previous session limit
- [BUG] Session limit reached far ahead of time compared to entire month of March.
- Resuming sessions from v2.1.89 in v2.1.90 causes excessive token usage (no prompt caching)
- [Bug] Unexpected High Resource Consumption from Plugin Disabling Commands
- [BUG] Authentication Key Lost on Claude Code Updates
- [Bug] Opus agent invocations charged despite available Fable usage credits
- $1,771 charged in 4 hours against a $25 spending limit
- [Bug] Security-guidance plugin bypasses subscription billing and uses expensive model without warning
- [BUG] Thinking stalls exceeding conversation cache TTL
- [Bug] /status reports costs for subscription plan when no billable usage occurred
- Sonnet[1m] requires extra-usage even on MAX 20x subscription
- You've hit your limit · resets 3pm (Asia/Seoul)
- You've hit your limit · resets 1pm (Europe/London)
- [Bug] Excessive token consumption for simple matplotlib plotting tasks
- [MODEL] Unacceptable Weekend Abandonment After Shipping Broken Release - Multiple Critical Failures
- [BUG] claude code subagent path-extract (Extract any file paths that this command reads or modifies) , use all bash output and not TRUNCATED, Causing a large amount of token fees
- [BUG] Claude Code using Sonnet 4.5 despite Haiku 4.5 being configured everywhere
- [BUG] Claude Code reporting I hit weekly usage limit
- Bug Report
- Claude AI usage limit reached|1749924000
- [BUG] Resuming a fork forfeits its prompt cache: deterministic miss after any tool loop, fork 5m vs parent 1h TTL, full-context rewrites while the prefix is live
- [Bug] Excessive token consumption during data processing tasks
- [BUG] Fable 5 / native-1M sessions silently clamp to a 200K context window on Max plan — undocumented, no user-facing notice, no recovery path
- Orphaned background --fork-session worker runs for hours and is respawned by the cc-daemon after kill (silent quota burn)
- Blocked by "Usage credits required for 1M context" with no in-session recovery path
- Sudden usage limit spike from 38% to 100% in 20-30 seconds on fresh sessions
- [BUG] Issue: /ultrareview returned empty findings 3 times in a row, all 3 free reviews consumed without usable output.
- /goal evaluator has no circuit breaker — unsatisfiable conditions burn unlimited tokens
- [Bug] Missing "Pay Now" button when purchasing API credits
- [BUG] CronCreate: lacks predicate-gated execution and context isolation, leading to large unintended costs on repeated trivial checks
- [BUG] Using tens of thousands of tokens for trivial work
- [Bug] Session limit consumption rate is unpredictable and inconsistent
- [BUG] consumed 2 of 3 free ultrareview runs, now i"m told i will be charged
- After just 500k tokens i get - You've hit your limit · resets 2am, and previous it was just 700k tokens. Before last weekend i could do millions.
- After my session was suspended last night due to rate...
- [BUG] uncacheable system prompt caused by `Primary working directory:` in git worktree
- [Bug] Malware check prompts causing rapid quota exhaustion and code analysis refusals
- [Bug] Claude Code process continues running despite being stopped, causing unexpected API usage
- [BUG] /usage and /status return 429 rate limit after single request
- You're out of extra usage · resets 1pm (UTC)
- [BUG] 10-12% usage spike when switching accounts and continuing task
- Proposed GitHub Issue Title: `/cost Command Reporting Unclear for Claude Max Subscribers` This title: - Directly identifies the specific feature/command - Highlights the core issue of reporting ambiguity - Uses technical terminology - Is concise and descr
- Usage limit can stop a session mid-operation and leave a half-applied state
- [BUG] Waking an idle agent fork (`subagent_type: "fork"`) forfeits its inherited prompt cache on every wake — `messages_changed`, `cache_read` pinned to a fixed boundary, not TTL
- Weekly quota pool: one −14% step between windows, then partial rebound (Max 20x) — earlier −43.5% figure retracted
- [Bug] Claude Fable 5 returns usage credit error while Claude 4.8 Opus works normally in both claude.ai and claude code
- [bug/feature] Protect against dangerous irresponsible humans? - You've hit your session limit · resets 12:50pm (Europe/London)
- [Bug] Session timeout mid-task causes incomplete execution and excessive token consumption
- [BUG] Agent fan-out pays ~47K uncached startup tokens per small task, causing multi-million-token usage
- You've hit your session limit · resets 1:50pm (Asia/Damascus)
- [BUG] Claude Code does not bound repeated identical tool calls
- [BUG] Async sub-agent completion can rebuild warm prompt cache, quickly burning usage limits in long-running sessions
- [BUG] /usage no longer shows limit consumed vs. limit (progress bar/percentage removed)
- [BUG] Massive dump of context caching issues, duplication of actions, and bloat
- [BUG] Claude Code Max x20 limits are too low compared to Codex Pro x20, and reliability has degraded
- /ultrareview returns empty findings paired with rate-limit error across 3 consecutive runs
- [BUG] /ultrareview failed TWICE
- ultrareview: free run consumed when server-side rate-limited (run crashed before producing findings)
- [BUG] ToolSearch unlock invalidates prompt-cache prefix mid-session
- You've hit your limit · resets 4:30pm (Asia/Calcutta)
- [BUG] Subagent 'Warmup' drains 99%+ of subscription input tokens (regression since 2026-04-09; ref #17457 closed NOT_PLANNED)
- [Bug] Triple token billing on context continuation and auto-compact
- [BUG] Claude Code auto-compaction runs indefinitely
- [BUG] /release-notes injects full changelog into conversation context (eats tokens)
- [Bug] Session Token Limit Consumed Entirely in Single Operation
- [Bug] Token usage calculation incorrect for *non-peak* hours with 2 small sessions
- [Bug] Anthropic API Error: Rate Limit Reached on conversation initialization
- [Bug] Weekly rate limit discrepancy between Claude Code CLI and dashboard UI
- [BUG] Task tool model parameter silently ignored for team agents - inherits parent model
- [BUG] JSONL logs record partial output_tokens from message_start events instead of final totals
- [BUG] Autcompact runs forever and consumes all credits
- [BUG]
- [Feature Request] Add API endpoint for retrieving weekly token usage statistics
- [BUG] System Reminder STILL transfering ENTIRE CONTENTS of Edited files costing massive tokens for no good reason
- [Bug] Unexpected API Credit Depletion Despite Max Plan Subscription
- [BUG] Claude should give the pro user the option to take the upcoming usage limit penalty immediately
- Credit balance is too low
- Token Consumption Spike During Ultrathink Command Execution
- [Bug] Synthesis agent fails on structured output cap, discarding deep research results and doubling token spend
- [BUG]
- [Bug] Anthropic API Error: 429 rate-limit errors persist despite available quota on Max plan
- [BUG] Fixed per-session token overhead increased ~75% (30k→52k) mid-day on 2026-06-16, independent of context window usage or content
- Usage limit discrepancy: Claude Code blocked at "100% used" while claude.ai Settings shows "44% used" on same Pro account
- [BUG] 437 subagents is probably too many
- [Bug] Agent workflow resumes rerun completed tasks instead of resuming from checkpoint
- goal, eternal loop with exit hooks and potentially catastrophic outcome
- [BUG] Limits became stricter since last update (but I still use Opus 4.8, not Fable 5)
- [Bug] Feedback loops cause excessive token consumption with unresponsive background jobs
- [BUG] Empty chat, single 'hi' message consumes 47.7k tokens
- /model command does not persist: Claude Code reverts to Sonnet 1M context (paid credits) despite selecting default Sonnet 4.6
- [BUG] `/ultrareview` crashes before producing findings — 2 consecutive failures, same branch
- [Bug] Session cost significantly higher than expected; parallel agent forking causing excessive API usage
- [Bug] Excessive token consumption during framework analysis and code generation
- You're out of extra usage · resets 6pm (America/Toronto)
- [Bug] Anthropic API Error: Rate Limited - Credit Consumed Without Review Generated
- /usage command lost its progress bar — now opens settings dialog instead