ChatGPT and Codex quota merge (December 2026): what OpenAI has and has not said
Shared by Pavel · Updated 2026-11-16
As of 2026-11-16. Every source below was read on 2026-11-16.
Labels: [vendor] = OpenAI's own page or message; [measured: who] = a user's own numbers, unverified by OpenAI; [opinion] = a user's view or demand; [third-party] = a non-OpenAI write-up.
On the OpenAI forum, "staff" means the OpenAI staff flag (OpenAI_Support). GitHub shows no staff flag, only an association label, so on GitHub "no OpenAI reply" means no comment that identifies itself as OpenAI's.
The questions people are asking
- What are the actual per-plan allowances (Plus, Pro, Business) after the merge, and how do they compare before vs after? OpenAI has not said. The help article gives no numbers. Asked in github.com/openai/codex/issues/58655 (2026-11-02, asd454314633-byte), #58530 (guisba, WilliamsSum and others, 11-01), #59079 (RoyAgiv, 11-03); forum t/1411466 (PZA 11-01, williamssum546, yamauchi.JP 11-02 and about 20 more), t/1411595.
- How many allowance units does each model and reasoning level cost in the shared pool? Partly answered. The Codex pricing page lists credit rates per million tokens and local-message ranges per 5 hours for Plus and Business. OpenAI has not said how Chat turns map to the same units, or how subscription allowance (not credits) is weighted. Asked in #58655, #58530, #61783 (cnoize35, 11-16); forum t/1411466.
- Will Chat Medium, High and Extra High reasoning draw from the shared allowance, or only Work/Codex? Partly answered. Help article: Instant does not draw; usage depends on model and task. It does not name High or Extra High chat. Staff reply 11-12: "You can use Instant without drawing from the shared allowance". No explicit statement for each reasoning level. Asked in forum t/1411466 (SummerFlowerouo 11-01, brandon.santiago.cha 11-02, Lune18 11-04), t/1415546 (yai 11-12); #58530.
- Will the 5-hour and weekly reset windows stay as they are, and how will reset times show for the combined pool? Partly answered. Help article: combined remaining usage and reset times will be visible in Settings > Usage; "we expect most users won't need to wait for their usage to reset more often than they do now". OpenAI has not said whether the 5-hour/weekly structure stays, or how it interacts with Pro's "no five-hour limit". Asked in forum t/1411466 (craznar 11-02, RoyAgiv); #59079; #60835.
- On what exact date does the change take effect? Partly answered. Help article: "Starting in December". OpenAI email (as reposted on Hacker News): "in the coming weeks". OpenAI has not given a day, or said whether rollout is staged or A/B tested. Asked in forum t/1411466 (craznar 11-04: "I thought it started in December?"); #58655.
- Can I keep separate pools, opt out, keep a protected reserve for Chat, or choose a classic plan? OpenAI has not said. Asked in #58655 (xhu81380-star 11-01), #59079 (bbj94, A/B proposal), #58530; forum t/1411466 (Lufiadani 11-02, AxelHu 11-11, many others), t/1416490.
- Business seats: is the shared allowance per user or per workspace, and how do credits fit? OpenAI has not said. The pricing page says Business plans have per-seat limits and credits; nothing on the merge. Asked in #60835 (giuliomarinelli 11-11; RoyAgiv 11-12); forum t/1411466 (wadofmeat, via #58530).
- Do conversations with a dot (and dot background work) draw from the shared allowance? OpenAI has not said. #60585 (MarkTian-Long, 2026-11-10) has 0 comments.
- Will Plus users get more Work/Codex capacity because of the merge, or is it only Chat High/Medium drawing from the same pool? OpenAI has not said. Asked in forum t/1411466 (yai 11-02, Lune18 11-04); #58655.
- Are Enterprise, Edu, Free and Go in scope? Answered from vendor text: the shared allowance covers "Plus, Pro, and Business". Pricing page: Enterprise/Edu without flexible pricing have the same per-seat limits as Plus for most features. Free and Go are not mentioned in the help article. OpenAI has not said how Enterprise/Edu are treated after December.
- How do I carry context from brainstorming to production, and what can I use to brainstorm without using quota? Answered by OpenAI_Support (staff) on 2026-11-12 in forum t/1415546 (asked by yai, who mentioned a "6.2-Sol-Instant" model): use Instant; use a Project, then a Work chat; in the desktop app, add an existing ChatGPT chat to a Codex chat. On "6.2-Sol-Instant", staff found no official documentation.
- Will I see usage per product, model and task, and get a warning before expensive runs? OpenAI has not said. The help article promises combined remaining usage and reset times only. Asked in #58655 (asd454314633-byte 11-02), forum t/1411466 (PZA 11-01, yai 11-01).
- If entitlements materially fall, is there a refund or exit option? Would the claimed October 29 doubling (a user report, unconfirmed) carry over? OpenAI has not said. Asked in #58655; forum t/1411466 (Drexl via #58530 11-01, jrtserrano 11-04, HWG39989 11-08, annual Business).
- Is the "October 29 doubling" of the Work/Codex allowance for Pro 100/200/500 real, and does it offset the merge? No readable primary OpenAI source confirms it (see What each plan gets and Open questions). Asked in forum t/1411466 (ShafiqB 11-01, honandy91 11-04).
- Which plan and usage comparison should I use to choose between Plus and Pro 100/200 (model-aware)? OpenAI has not said. #61783 (cnoize35, 2026-11-16) has 0 comments.
Related but different live reports, about how the usage meter behaves now rather than the merge: #59690 (CoreyHayward), #60530 (eleisenberger, 11 comments), #61423 (battle-mage), #59639 (sofiapaixao-tv), #59882 (cdenn, unattributed "Unknown" usage), #58925 (workspace members cannot see remaining allowance). No staff comments on any of these.
What changes and when
- [vendor] Help center article "Understanding ChatGPT and Codex usage limits"; the page says "Updated: 15 days ago" (about 2026-11-01): "Starting in December, Plus, Pro, and Business plans' usage allowance will be shared across ChatGPT and Codex as we bring the agentic capabilities of Work into the main ChatGPT experience." https://help.openai.com/en/articles/20001625-understanding-chatgpt-and-codex-usage-limits
- [vendor] Same page: usage "will be measured in tokens"; consumption depends on model and task; "More advanced models and complex tasks generally use more of your allowance. Instant chat will not draw from your usage."
- [vendor] Same page: "We expect most users won't need to wait for their usage to reset more often than they do now." Combined remaining usage and reset times in Settings > Usage. At the limit: wait for reset, upgrade, or buy credits for eligible features.
- [third-party, reposted email text] Hacker News story "A simpler way to put ChatGPT to work (A shared allowance for ChatGPT and Codex)", submitted 2026-11-01 by HN user daneel951, quoting an OpenAI email: "In the coming weeks, you'll be able to delegate work directly to ChatGPT ... without switching to the Work tab"; "your use of ChatGPT and Codex will draw from the same allowance in the coming weeks"; "Unlimited Instant is included in your plan and will not draw from your usage." https://news.ycombinator.com/item?id=50324512 . Not verified against openai.com.
- Date conflict: the help article says "Starting in December"; the emailed text says "in the coming weeks". The same HN user posted a second email in a comment, subject "Chat and Work are coming together in one experience", saying a combined Chat/Work experience will be tested with a small group "in the coming weeks" and repeating the same-allowance and Unlimited Instant lines. Different accounts may have received different emails.
- [vendor] The Codex pricing page already says "ChatGPT Work and Codex share usage. ChatGPT Work usage inside ChatGPT uses the same pricing, credits, and usage limits as Codex." (https://developers.openai.com/codex/pricing, redirects to https://learn.chatgpt.com/docs/pricing/). So Work and Codex are already one pool today; what the December change adds is regular ChatGPT (non-Instant) chat.
- [vendor, absence] As of 2026-11-16, the Codex pricing page, the Codex/ChatGPT changelog (https://learn.chatgpt.com/docs/changelog/, entries through 2026-11-13, CLI 0.169.0), the What's new page, and openai/codex CHANGELOG.md (points to GitHub releases; latest release seen rust-v0.171.0-alpha.8, 2026-11-16) contain no entry about the shared allowance or a date.
- [vendor, absence] A staff reply on 2026-11-12 on the forum links the help article as "the December usage update", adding no date or numbers.
What each plan gets
[vendor] unless stated. Source: Codex pricing page (https://developers.openai.com/codex/pricing -> learn.chatgpt.com/docs/pricing), as of 2026-11-16. These are current (pre-merge) Codex numbers; no post-merge numbers are published.
- Plus ($20/month on the pricing page): local messages per 5-hour window: GPT-6 Astra 5-45, GPT-6.1 Sol 15-160, GPT-6.1 Terra 35-285, GPT-6 Sol 15-150, GPT-6 Luna 350-3,000. Standard Business shows the same ranges. "These estimates are not fixed message limits." Weekly limits may also apply.
- Pro: plans at $100, $200 and $500 per month. "Pro plans currently have no five-hour limit." The page defers to the dashboard for limits; no per-model Pro numbers on the page. Ultrafast on Pro $500.
- Business: $20 per user per month billed annually (2+ users), $25 monthly; credits can extend usage.
- Enterprise/Edu: with flexible pricing, no fixed rate limits, usage scales with credits; without it, "the same per-seat usage limits as Plus for most features".
- Free and Go: Codex included; Luna in the desktop app subject to rollout.
- Shared allowance scope per the help article: Plus, Pro, Business. Nothing per plan on how big the shared pool is.
- Other agentic features already pool: "Usage limits are shared with other agentic features once pricing for those features is effective. This currently includes ChatGPT for Excel on Plus and Pro."
- [opinion, unverified] User reports only: forum users ShafiqB (11-01, "OpenAI has announced a doubling") and honandy91 (11-04, "In its October 29 update, OpenAI doubled the Work/Codex allowance for Pro 100 and Pro 500, and for Pro 200 users who had the lower allowance", which they call their interpretation) cite an "October 29 update". No readable OpenAI page confirms it, its scope or its link to the merge. Another user (fidaner, 11-02) claims instead that OpenAI cut Pro 200 allowances on "Sep 29"; also unconfirmed. These reports conflict or are unverified; neither is a fact.
How usage is metered and shown
- [vendor] Tokens: prompt, files, history, tool results and response all count (pricing page FAQ "What are tokens and credits?").
- [vendor] Credit rates per 1M tokens, Standard speed (input / cached input / output): GPT-6 Astra 250 / 25 / 1,250; GPT-6.1 Sol 50 / 2.5 / 250; GPT-6.1 Terra 25 / 1.25 / 125; GPT-6 Sol 50 / 5 / 250; GPT-6 Luna 2.5 / 0.25 / 12.5; GPT-5.6 Sol 100 / 10 / 500. The page says "Credit prices alone don't determine included subscription usage; check your usage dashboard." No separate cache-write charge in credit billing.
- [vendor] Fast mode uses 2.5x included subscription usage (2x purchased credits); Astra Ultrafast and Sol Ultrafast 8x included (6x credits), relative to Standard. Image generation uses included limits about 3-5x faster. Desktop voice: $0.05 per minute of Codex usage budget (1.25 credits per minute on credit plans).
- [vendor] "Local messages and cloud chats share your plan's usage allowance."
- [vendor] At the limit mid-turn: the agent can continue the active turn "subject to fair use limits". Plus and Pro can buy credits; Business/Enterprise/Edu with flexible pricing can buy workspace credits.
- [vendor] Where it is shown: the usage dashboard (https://chatgpt.com/codex/settings/usage) and /status in the CLI; after the merge, Settings > Usage with combined remaining usage and reset times (help article). Staff reply 2026-11-13 (forum t/1404599): Codex shows 5-hour and weekly separately in Settings > Usage, each with a reset time; usage continues only while both have capacity.
- [opinion: forum users PsyPrime and yai, https://community.openai.com/t/1411466, 2026-11-01] PsyPrime points to Settings > Usage > Analytics and Settings > Profile as the stats views; yai says they show estimates and can't tie credits to requests. Not a measurement; unverified.
- [vendor, gap] Nothing published on how a regular Chat High/Extra High turn converts to allowance units, or whether Settings > Usage will break usage out by Chat vs Codex.
What users measured
None of these measure the merge, which has not launched. They show how the existing meter behaves, which is what the shared pool will inherit. All are [measured: named user], unverified by OpenAI, with no staff response on any.
- CoreyHayward, github.com/openai/codex/issues/59690, 2026-11-05, Business plan, codex-cli 0.166.0: from the morning of 5 Nov, GPT-6.1 Terra calls used about 15x more of the 5-hour limit per input token than on 1-4 Nov; cached input appeared to be billed at the full input price; 74 Terra calls used 34 points of the weekly limit on 5 Nov. Update 2026-11-06 by the same author: "this looks fixed", after a table comparing Business 1-4 Nov (1,109 calls, 0.32% of the 5-hour limit per call) with later data. Other users (11-07 to 11-13) report similar Terra/Sol drain afterwards.
- eleisenberger, #60530, 2026-11-10, GPT-6.1 Sol: weekly usage 100.2% for Nov 2-9; one conversation used 88.96% of the weekly limit; the dashboard says some usage information is unavailable. A comment thread of 11 with a dozen "same here" reports (Plus and others), e.g. a 10-page .docx analysis dropping 5-hour remaining from about 90% to 53% (Pdelgadoo, 11-12), and the 5-hour limit gone in 20-30 minutes (MegaDroid7, CloudBudgiePatrick, 11-13).
- battle-mage, #61423, 2026-11-14, Plus, Windows: two 5-hour windows consumed 31% of the weekly limit; six windows nearly 100%.
- sofiapaixao-tv, #59639, 2026-11-05, Business Premium: one GPT-6 Astra Ultra repo-wide review consumed about 37% of the weekly allowance; 8 percentage points in about an hour on Sol High/Xhigh.
- charlie2233, comment on #58530, Pro, 2026-11-02: Codex weekly allowance exhausted in a week, with the profile showing about 1.1B tokens of activity (the poster notes the total isn't necessarily what's counted).
- [opinion] Comments on #58530 from several users: "I regularly exhaust my Codex allowance but never hit my advanced Chat limits" (mcburgs, 2026-11-13), i.e. heavy Codex users expect the pooled meter to be consumed from the coding side; light Chat-only users (tamaradenmeade, HodgePodge789, shuangjinggu) expect the opposite. These are opinions about the expected effect, not measurements.
- Suggestion, not a measurement: Zyph3rKeaton (#58825, 11-02) proposed a ledger of identical tasks, usage and reset timestamps before vs after December for a clean comparison. Nobody has published one.
- No one has published a before/after comparison of the shared pool, because it is not live. A forum user (honandy91, 11-04) says "the weekly quota now lasts longer" for them and agrees it is an interpretation, not an OpenAI statement.
Open questions with no vendor answer
As of 2026-11-16:
- Size of the shared allowance per plan (Plus, Pro 100/200/500, Business seat), and any before/after comparison.
- How Chat turns at Medium, High and Extra High map to allowance units; whether any cap per turn or a reserve exists.
- Whether weekly and 5-hour windows both apply to the pool; Pro's current "no five-hour limit".
- Exact start date ("December" in the help article vs "coming weeks" in the emailed text); whether rollout is staged or A/B tested.
- Dots: whether dot conversations and dot background work draw from the pool (issue #60585, unanswered).
- Business: per-seat vs workspace-pooled allowance; credit handling.
- Whether a user can opt out, keep a separate Chat allowance, or choose a classic plan.
- A breakdown view by product, model and task, and any warning before an expensive run.
- Treatment of existing annual Business subscribers and any refund or exit path if effective capacity falls (a policy question; no legal conclusion is drawn here).
- The "October 29" Pro doubling: no primary page found that confirms it, its scope, or how it relates to the merge.
- Net effect for Plus: does Work/Codex capacity grow when Chat High draws from the same pool?
- Whether GPT-6.1 Sol and Terra, which the pricing page says are available in Work and Codex "not Chat", remain outside Chat after the merge (the page says Sol is not available in Chat today).
Sources
All read 2026-11-16.
- [vendor] https://help.openai.com/en/articles/20001625-understanding-chatgpt-and-codex-usage-limits
- [vendor] https://developers.openai.com/codex/pricing (redirects to https://learn.chatgpt.com/docs/pricing/)
- [vendor] https://learn.chatgpt.com/docs/changelog/ (via https://developers.openai.com/codex/changelog), https://learn.chatgpt.com/docs/whats-new
- [vendor] https://raw.githubusercontent.com/openai/codex/main/CHANGELOG.md (stub) and https://github.com/openai/codex/releases
- [vendor, staff replies] https://community.openai.com/t/1415546 (OpenAI_Support, 2026-11-12); https://community.openai.com/t/1404599 (OpenAI_Support, 2026-11-13, Codex 5-hour vs weekly display)
- [third-party] https://news.ycombinator.com/item?id=50324512 (email text reposted by daneel951)
- GitHub: https://github.com/openai/codex/issues/58655, /58530, /58825, /59079, /59311, /60585, /60835, /61783; for metering /59690, /60530, /61423, /59639
- Forum: https://community.openai.com/t/1411466, /1411595, /1416490, /1416317, /1416526, /1409401 (Codex Rate Limit Discussion Thread; no staff post since 11-02)
- Not available: openai.com/chatgpt/pricing, the OpenAI announcement email on openai.com.