October 3 access correction

Vendor terms checked October 3, 2026: OpenAI’s Pro FAQ now permits new Pro $200 purchases. Eligible existing subscribers keep the previous allowance through October 29, then receive a lower allowance at the same $200/month. This supersedes the September purchase-pause and 30-day return guidance below. Detailed cohort eligibility and same-account return conditions belong in the launch/offers owner. This is a source check, not a checkout or account-entitlement test; older caching, reset and incident evidence retains its dates.

Historical author report, September 14, 2026: The AIHackers author reported verifying that Pro 20x was unavailable for new purchase then. This account observation supplemented the provider notice; it did not establish every account’s eligibility and is superseded by the October correction.

October 3 local network and cache observations

Site-owned, one Debian 13 environment: the retained October 3 validation receipt records Codex CLI/daemon 0.160.0 and gh 2.102.0. A fresh app-server executed a small authenticated GitHub request and the CI-log command in its configured default sandbox, without a command sandbox override or full-access retry. The log contained 114,048 bytes, SHA-256 5cfc0576236488212247b7461a4768ebb927bc3ca0a25f9509603455e85dcb24. An uncached download was also tested during the diagnosis.

The permission profile extended :workspace with an enforcing network proxy and six exact allowed hosts: api.github.com, github.com, objects.githubusercontent.com, raw.githubusercontent.com, codeload.github.com, and the observed log destination results-receiver.actions.githubusercontent.com. An unlisted HTTPS destination received proxy CONNECT 403; protected mounts remained read-only. These hosts describe the tested environment, not every GitHub operation or Enterprise backend. OpenAI permissions documentation independently establishes the separation between local filesystem and network controls.

The old gh 2.46.0 returned empty rendered output in the comparison; the newer client returned the log. No minimum fixed version was established. An inherited cache path outside writable roots also failed. GitHub documents archive-to-job log fallback, but this observation does not prove one universal parser defect. Saved browser settings changed while the running MCP process retained its old launcher environment; Chrome’s extension/native bridge were absent. Browser operation and IAB availability remain unverified. Demand, prevalence and time saved are not-measured.

Separate cache observation: at review time, the latest completed local root session using only gpt-6.1-sol reported these cumulative client counters:

Client fieldTokens
input_tokens8,540,577
cached_input_tokens8,325,760
output_tokens52,341
Cached share of input97.5%

The session used Codex 0.160.0 and completed October 3 at 10:59:53 UTC. The calculation is 8,325,760 / 8,540,577, rounded to one decimal percent; cumulative events were not summed repeatedly. This deliberately selected local example demonstrates high token reuse can occur. It does not estimate average user behavior, request-level hit rate, API cache-write billing, subscription drawdown or accepted-result cost. No API invoice, per-request pricing-tier breakdown, paid cache experiment or controlled quality evaluation was collected; CAR remains not-run. Raw conversation text and credentials are excluded from the research record.

Current vendor rates and cache mechanics are owned by the Sol 6.1 pricing guide and prompt-caching explainer. Auto-review is a reviewer swap, not the cause of cache reuse or a permission grant. The Codex guide connects these observations to practical diagnostics.

October 3 polling evidence check

Community-observed: Codex issue #13733, opened March 6, reports repeated model turns during background process checks. Issue #31935, opened July 9, reports a 60-second blocking-wait instruction; #32188, opened July 10, requests wake-on-exit support. A Reddit thread opened for this review describes similar checking loops and inconsistent results from local instructions. Its exact publication timestamp was not established from the fetched page. Reporter token/quota estimates are not audited AIHackers results.

Vendor-documented: the current configuration reference lists a 300,000 ms default maximum background-terminal poll window. That is a wait ceiling, not a universal model polling cadence. gh run watch separately defaults to a three-second GitHub refresh. Its --interval controls that subprocess; it does not control model re-entry. A completion feature request is not a supported portable tool flag.

Recommendation, not a measured saving: retain one watcher/session, use supported completion events or appropriately long waits within current tool/instruction limits, inspect actionable changes, and verify the final result. The worked guide supplies scoped instructions. Prevalence, avoidable inference counts, cash/quota savings and a causal before/after comparison remain not-measured. Current-source archives for these additions are archive-pending.

September 27 access and reset update

Historical vendor check, September 27: the then-reviewed Pro FAQ retained the new-subscription and upgrade pause, with existing Pro $200 renewals continuing and Pro $100 available. It permitted an eligible one-time return within 30 days after Pro $200 access ended, requiring access on September 10 or access that ended in the preceding 30 days, and an unused return. That superseded our September 14 blanket no-return statement. The October 3 correction above replaces these access conditions; they are not current purchase advice.

Tibo announced a paid Codex/Work incident reset on September 26 UTC, then reported propagation complete. This is vendor event evidence, not proof from every account or a standing reset schedule. The launch offers guide separates that incident from other grants and now records released DevDay terms. New source archives: archive-pending. Older caching and usage findings below retain their original review dates.

September 14 Pro 20x access check

Historical check, September 14, 2026: OpenAI’s Pro terms paused new Pro $200/20x subscriptions and upgrades from September 10. Existing renewals continued; Pro $100 remained available. At that check, the terms did not offer the eligible one-time return now described above. Archive: archive-pending.

Our September recommendation preserves its dated retention advice with an October access correction. It does not promise monthly capacity. This historical check updates access only; older quota observations below retain their dates.

September 27 current reset mechanics

Vendor-verified, checked September 27, 2026: OpenAI’s banked Codex reset help page says a full banked reset refreshes both the five-hour and weekly Codex usage windows and moves the weekly reset date. Its example says that a Friday weekly reset followed by use of a full reset on Tuesday moves the next weekly reset to around the following Tuesday; the original Friday reset does not also occur. Eligibility, affected windows, delivery timing, and expiry vary by offer, plan, workspace, region, and account. The same page records banked resets for eligible Plus, Pro, and Business accounts on September 3 and 4, 2026, while stating that future resets are not guaranteed.

OpenAI separately documents an instant paid weekly Work and Codex reset for eligible personal Plus and Pro accounts. It restores both five-hour and weekly usage, pulls the normal weekly allowance forward instead of adding a separate entitlement, applies immediately, cannot be saved, and starts a new weekly period with the first Work or Codex request after the reset. The next automatic reset is seven days after that first request. The paid reset is separate from banked resets and usage credits; availability varies by account and the current help page excludes Free, Go, Business, Enterprise, and Edu plans. The help page does not establish universal pricing.

These current primary-doc findings supersede the August-only reset observations below. Compaction, caching, incident, and other historical rows retain their original evidence dates.

Verdict

Agent loops can consume more tokens than a single response, and compaction can be followed by costly rereads. The evidence does not establish a universal Codex inflation multiplier, a five-file/50K context rule, or an OpenCode efficiency advantage.

Use the labels below precisely:

  • validated — current primary documentation or reproducible source evidence supports the claim.
  • version-specific — a report concerns named software versions and should not be generalized.
  • account-level — evidence describes one account, plan, or usage meter.
  • community-observed — dated user reports exist without first-party confirmation of scope or cause.
  • unresolved — evidence is incomplete or conflicting.
  • speculative — a labeled hypothesis or forecast is not established provider behavior or intent.
  • incorrect — current evidence contradicts the claim or its attribution.

Archive state is separate. validated means a concrete replayable capture exists; archive-pending means it does not. A confidence label never manufactures archive proof.

Claim ledger

ClaimStatusFindingArchive
Agent and subagent loops add model workvalidatedCodex documentation says subagents do their own model/tool work and use more tokens than comparable single-agent runs. No universal multiplier is documented.archive-pending
“Codex inflates tokens by 40–80%”incorrectOpenAI’s 40–80% figure is an internal-test improvement in cache utilization for Responses versus Chat Completions. It is not Codex inflation or waste.validated
Codex automatically includes five recently edited files or 50K tokensincorrectCurrent Codex documentation and searches of the pinned 2026-07-26 source revision do not document this rule. Treat it as incorrect or misattributed unless a versioned source is produced.archive-pending
OpenCode is more token-efficient than CodexunresolvedHarness, model, provider adapter, prompt shape, tools, compaction, retries, and task success all affect usage. No controlled same-task dataset establishes a general winner.n/a
Prompt caching makes prior context freeincorrectCaching discounts eligible exact-prefix reuse. Cached input still occupies context, and stateful Responses chains still bill prior input.validated
GPT-5.6 cache writes cost more than uncached inputvalidatedOpenAI documents cache writes at 1.25× uncached input for GPT-5.6 and later families, with reads billed at the cached-input rate.validated
Codex v0.118 lowered the compaction thresholdversion-specificIssue #16812 alleged a regression, but a maintainer found no threshold or estimator change and the reporter’s controlled test reversed the original result. The reporter closed it as not a bug.validated
A July 24 Desktop session looped through compaction and rereadscommunity-observedOpen issue #35226 records one GPT-5.6 Sol/Desktop 26.721.31836 session. Its token, quota, and causal estimates are reporter-supplied; no maintainer diagnosis is published.validated
The June 26 usage incident proved service-wide token inflationincorrectOpenAI attributed some reports to fraud/abuse systems incorrectly rate-limiting certain accounts, said impact appeared limited, and reported no broader Codex degradation.validated
July 25 resets compensated for caching defectsincorrectA first-party announcement says the reset reached all Codex and ChatGPT Work users after an almost global outage. It does not attribute the reset to caching or compaction.archive-pending
July 28 through August 13 broad resets fixed caching defectsincorrectDirect announcements establish reset scope but do not identify prompt caching or compaction as the cause. Efficiency, product, calendar, and milestone framing do not establish causality.archive-pending
Codex has a documented and universally active five-hour windowunresolvedOpenAI’s current banked-reset and paid-weekly-reset help pages explicitly say a full reset restores both five-hour and weekly windows. That confirms reset mechanics, not universal enforcement, availability, or account state; account behavior may vary.pricing validated; current help pages checked 2026-09-27, archive archive-pending
The August 8, 11, and 13 announcements created a permanent reset scheduleincorrectThe posts establish three resets, not a standing entitlement. The August 12 post says the earlier every-million promise ended at 10M; the 15M reset is a discretionary milestone event.archive-pending
A new broad banked-reset grant was announced after July 13vendor-confirmed, September 27 reviewOpenAI’s September 22 launch announcement relays a banked-reset grant for Plus, Pro, and Business. Account receipt and expiry still need an account check.archive-pending
Weekly capacity plus discretionary resets is the near-term equilibriumspeculativeHistorical August forecast, not a provider commitment. The September 22 broad banked grant changes its evidence base; the earlier no-later-grant finding is superseded.n/a
Purchased credits refill five-hour or weekly countersincorrectOpenAI says available credits let work continue after included limits. It does not describe credits as restarting the shared five-hour window or refilling a weekly counter.validated
Eligible personal Plus and Pro accounts can purchase a full weekly resetvalidatedOpenAI documents an immediate paid reset from Usage settings or an in-app weekly-limit offer. Availability varies by account; the current page excludes Free, Go, Business, Enterprise, and Edu plans and does not establish universal pricing.current help page checked 2026-09-27; archive-pending
Purchased resets cost $8 for Plus and $80 for Pro 20x universallyincorrectThose August observations were account-specific prices, not a universal price sheet. At the August review, Pro 5x coverage, taxes, refunds, and eligibility were unresolved. OpenAI’s September 27 help page now documents eligible personal Plus and Pro accounts, excludes Business, and directs users to the account checkout price; universal pricing remains unpublished.validated community captures; current help page checked 2026-09-27, archive archive-pending
A purchased reset stacks a bonus week without moving scheduled recoveryincorrectOpenAI says the paid reset pulls the normal weekly allowance forward instead of adding a separate entitlement. It applies immediately, and the next automatic weekly reset is seven days after the first Work or Codex request after the reset.current help page checked 2026-09-27; archive-pending
A purchased reset is the same as a banked reset or usage creditincorrectOpenAI describes a paid reset as an immediate purchase, a banked reset as an account-held benefit for later use, and credits as a separate pay-as-you-go balance. These mechanisms have different timing and accounting.current help pages checked 2026-09-27; archive-pending
Paid resets replaced free promotional resetsunresolvedThe bounded evidence shows a paid control and separate historical provider-wide or banked grants. It does not establish that OpenAI ended or replaced future promotional resets.n/a
Sol allowance now lasts 18% longer for everyoneincorrectOpenAI projected around 18% longer allowance during typical Sol use after improvements. This is provider-reported, workload-dependent, and not an AIHackers benchmark or universal account result.archive-pending
A banked reset count applies to every planaccount-levelBanked resets and their visible count depend on eligibility, promotion, plan, workspace, and account state.archive-pending
One referral reward or expiry applies universallyincorrectCurrent ChatGPT Desktop terms make the benefit, eligible experience, qualifying action, cap, cooldown, redemption window, and expiry offer-specific. Referral credits are not API credits unless the applicable offer explicitly says otherwise. See the extra Codex usage referral guide; AIHackers publishes no OpenAI referral link.validated; prior Codex-titled revision captured 2026-07-10

What current Codex evidence actually says

The current Codex manual describes /compact as a way to summarize visible chat and free context, and /status as the place to inspect token usage and remaining capacity. Current source exposes configurable context, auto-compaction, and tool-output token limits.

At pinned commit 61a4488, source search finds compaction thresholds and context accounting. Searches for “recently edited,” “recent_files,” and “50K” do not establish a five-file preload contract. Absence from a source search is not proof that no private or future surface can behave that way; it is enough to reject the claim as documented current Codex behavior.

Read the compaction reports narrowly

Closed April report: version-specific

Issue #16812 compared v0.116 with v0.118 and initially attributed higher usage to more frequent compaction. A maintainer reported no compaction-threshold or token-estimator change. The reporter then ran an identical-prompt test in which v0.116 compacted three times and v0.118 twice, contradicting the original regression theory, and closed the issue as not a bug.

The report remains useful evidence that long, noisy tasks can reread material after compaction. It is not evidence of a confirmed v0.118 defect or a universal multiplier.

Open July report: community-observed

Issue #35226, opened July 24, describes one Desktop session repeatedly rereading files after automatic compaction near a full context window. The issue is open as of July 26. Its estimated 10–15% paid-usage impact is a user estimate, not an audited refund amount, root cause, or platform-wide rate.

Plan-change reports: historical evidence

The reset chronology’s plan-change section preserves the April 29, May 8, and July 5 forum threads and the August 15 one-account follow-up formerly discussed in the cost playbook.

ClaimEvidence statusBoundaryArchive state
Some users reported depleted allowance after payment or a plan changecommunity-observedDated, account-specific, non-representative reports; no established bug or reset mechanismOriginal thread links retained in chronology; no new capture claimed
An August 15 reply describes Free/Plus/refund/Pro transitionscommunity-observedOne account’s sequence, not a workaround or purchase recommendationPreviously validated August 24 capture retained; no new archive validation claimed
Changing plans guarantees a fresh Codex usage windowunresolvedReviewed provider documentation does not establish this entitlement; an API route has separate authentication and billingarchive-pending for current documentation

Incident and reset chronology

  • June 26, investigating: OpenAI investigated reports of faster-than-expected Codex usage-limit consumption.
  • June 26, monitoring: OpenAI attributed some reports to incorrect fraud/abuse rate limiting of certain accounts, described the impact as limited, and said it had not observed broader degradation.
  • June 29, resolved: OpenAI marked the incident recovered.
  • July 25: OpenAI’s product lead announced a reset for all Codex and ChatGPT Work users after an almost global outage.
  • July 28: he announced a reset for all paid Codex and ChatGPT Work users.
  • July 29: he announced a reset for all Codex and ChatGPT Work users, projected around 18% longer typical Sol allowance, and said the temporarily paused five-hour limit would be restored July 30.
  • August 1: he announced another reset for Codex and ChatGPT Work with efficiency and Luna messaging; the post does not identify the efficiency work as the reset’s cause.
  • August 8: he announced a reset for all paid ChatGPT Work and Codex users.
  • August 11: he announced another reset for all paid ChatGPT Work and Codex users.
  • August 12: he said the earlier every-million promise ended at 10M and teased a surprise after growth passed that milestone.
  • August 13 at 01:01 UTC: he said Codex crossed 15M active users and announced a reset for “everyone.”
  • August 11 onward: dated account reports showed purchased full-reset controls at $8 on some Plus accounts and $80 on some Pro 20x accounts; availability remained uneven.

Keep scheduled limit recovery, provider-wide hard resets, incident remediation, manually applied banked resets, purchased full resets, and purchased usage credits in separate buckets. Reported purchased resets start a new seven-day window; credits continue work after included limits and are not documented as refilling counters. No cited announcement ties these resets to caching defects.

OpenAI’s June 11 release notes say banked resets are usable for 30 days after grant. A July 12/13 direct announcement described a 500,000-user grant and the first broad banked grant for the 7M milestone. Dated reports of grants disappearing around August 12–13 are community-observed; they do not prove universal expiry behavior, notification behavior, or formal withdrawal.

Reproducible comparison standard

To compare Codex with OpenCode, hold constant:

  1. model and provider;
  2. repository commit and clean starting state;
  3. prompt and project instructions;
  4. tools, MCP schemas, permissions, and network access;
  5. reasoning effort, timeout, retries, and subagent budget; and
  6. tests and acceptance criteria.

Record input, cache-write, cache-read, output/reasoning, tool calls, compactions, rereads, human review, and whether the result was accepted. Compare cost per accepted result, not raw tokens from mismatched tasks.

Sources

Global cache/source/issue review: July 26, 2026. Reset, pricing, paid-reset, credit, banking, and five-hour rows were rechecked August 20; unchanged compaction rows retain their earlier evidence dates.