11-02-2026, 03:27 AM
OpenAI is set to overhaul how paid subscribers access its products. Beginning in December, ChatGPT and Codex will share a single usage pool across Plus, Pro, and Business tiers, with consumption measured in tokens rather than message counts. The change, detailed in a help-center document, ends the era of separate allowances for chatbot conversations and coding sessions and closes a period of frequent promotional resets that Codex users enjoyed during a 28-day shipping campaign.
What the Shared Pool Means for Subscribers
Under the new system, Plus, Pro, and Business subscribers will draw from one combined balance across ChatGPT and Codex. Instant chat, the fast-reply mode built for quick questions, is exempt and will not consume the shared allowance. Everything else is charged according to the selected model and task complexity—heavier reasoning and tool use will drain the pool faster than a short prompt.
OpenAI framed the shift as a consequence of product consolidation, saying it is bringing the agentic capabilities of ChatGPT Work into the main ChatGPT experience. Work and Codex already shared a pool, so the adjustment mainly pulls everyday chat into that same accounting. Subscribers will be able to track their combined balance and reset schedule under Settings, then Usage. Once the pool runs dry, users can wait, move up a tier, or purchase credits for eligible features.
The timing is awkward for Plus subscribers in particular. Days before the pooling news, OpenAI doubled allowances for Pro 100, Pro 200, Pro 500, and Business Premium, and scrapped a planned cut to Pro 200 limits. Plus was left out of the increase. Those looking to secure a Cheap ChatGPT Plus Account before the December changes take effect may find competitive options through z2u.com.
The End of the Reset Era
The change did not come out of nowhere. On October 30, product tracker Tibor Blaho spotted a line on ChatGPT's checkout page warning that starting in December, plan usage would share a single ceiling across all ChatGPT and Codex features. Codex chief Thibault Sottiaux had said a day earlier that the team was working on the change but had not begun rolling it out.
The 28-day campaign produced a long list of improvements alongside the resets. Default speeds for GPT-6 Astra and GPT-6.1 Sol in subscription products rose about 50%. GPT-6.1 Sol's default context expanded from 272,000 to 500,000 tokens. Caching and request-routing work effectively stretched the same allowance further. GPT-6.1 Terra returned with markedly better writing and persona capabilities while consuming half the usage of 6.1 Sol. Resets, both global and banked, numbered more than ten across the cycle. On day 25 came the Pro and Business Premium doubling, accompanied by Sottiaux's line that the team was graduating "from The Reset Company to The Token Company."
New Device Limits Close a Workaround
A second restriction landed in the same window. A new help page on payment cards and device limits caps how many paid accounts a single device can use, covering Go, Plus, and the Pro 100, 200, and 500 tiers. Exceeding the cap can trigger a "Device limit reached" message at login or a mid-session logout, even after payment clears. Signing out of another account does not free a slot.
Business, Enterprise, Edu, and free accounts fall outside the limit. The rule arrives just as Codex added multi-account switching on mobile, closing off a workaround heavy users had relied on. The company points to credits purchased through the Codex app's usage and billing section as the sanctioned way to add capacity. For developers and researchers who need flexible access to older models, platforms like z2u.com offer ways to Buy OpenAI GPT-3 Access without committing to higher-tier subscriptions.
The device-limit documentation adds details that matter for anyone already holding several subscriptions. Purchases made before the restrictions took effect can continue under an existing-subscription exception, but canceling one does not guarantee a replacement will be approved. The limits attach to the payment card as well as the device, so a shared household card or a public device can trip them even when each person uses their own account. Buying credits or accepting a reset does not clear a card or device restriction.
Token Accounting Makes Compute Costs Visible
The arithmetic behind the shift is straightforward. Counting messages hides the true cost of a query: a weather question and a half-hour reasoning run looked identical on the meter, with OpenAI absorbing the difference. Token accounting turns a subscription into something closer to prepaid API capacity, with overage covered by credits and compute costs made visible to the user.
That pressure is real. After GPT-6 Astra launched in September, demand spiked hard enough that OpenAI paused new subscriptions to the $200 Pro tier. When sales reopened, Sottiaux acknowledged the new arrangement worked out to roughly half the API-equivalent spend of the old one. The frequent resets of the past month suggest the allowance is a valve the company can turn at will; now the method of adjustment shifts from handing out extra resets to uniform metering.
Competitive Pressure and Financial Stakes
The competitive backdrop sharpens the stakes. Anthropic more than doubled quarterly revenue to $11.6 billion in the second quarter from the prior period, while OpenAI's revenue rose about 18% to $6.7 billion, according to the Wall Street Journal. Anthropic posted a small operating profit for the quarter, in contrast to widening losses at OpenAI, which ran a $12.3 billion operating loss on that revenue and has filed paperwork for a listing expected next month.
For investors, the pooling change is a margin story as much as a product one. Moving to token metering shifts compute cost onto users who consume the most, a lever that matters when the company is preparing its own potential IPO as early as next year. OpenAI has secured roughly $11 billion in commitments for its latest round, against an initial $30 billion target that the company has since scaled back to as much as $20 billion. The round values it at $1.25 trillion, below the $1.4 trillion it sought, and the company plans to approach family offices and wealthy individuals for the remainder.
What Subscribers Should Know
For subscribers, the practical question is how large each tier's pool will be. OpenAI has not disclosed token allotments or the exact December start date, leaving users unable to calculate whether a combined pool will stretch further than the separate allowances it replaces. Light users who lean on Instant will likely notice little. Heavy users who spend the day reasoning in chat and the night running long coding jobs will now compete with themselves for one balance.
The era of effectively unlimited questioning narrows to a single mode. Instant chat remains free of the shared pool, but everything else now flows through one balance. For those exploring older model access alongside modern subscriptions, z2u.com provides a marketplace for GPT-3 Accounts and other OpenAI offerings, giving users flexible entry points as the pricing landscape shifts. Whether the combined pool will stretch further than the separate allowances it replaces remains the open question—one OpenAI has so far declined to answer with concrete numbers.
What the Shared Pool Means for Subscribers
Under the new system, Plus, Pro, and Business subscribers will draw from one combined balance across ChatGPT and Codex. Instant chat, the fast-reply mode built for quick questions, is exempt and will not consume the shared allowance. Everything else is charged according to the selected model and task complexity—heavier reasoning and tool use will drain the pool faster than a short prompt.
OpenAI framed the shift as a consequence of product consolidation, saying it is bringing the agentic capabilities of ChatGPT Work into the main ChatGPT experience. Work and Codex already shared a pool, so the adjustment mainly pulls everyday chat into that same accounting. Subscribers will be able to track their combined balance and reset schedule under Settings, then Usage. Once the pool runs dry, users can wait, move up a tier, or purchase credits for eligible features.
The timing is awkward for Plus subscribers in particular. Days before the pooling news, OpenAI doubled allowances for Pro 100, Pro 200, Pro 500, and Business Premium, and scrapped a planned cut to Pro 200 limits. Plus was left out of the increase. Those looking to secure a Cheap ChatGPT Plus Account before the December changes take effect may find competitive options through z2u.com.
The End of the Reset Era
The change did not come out of nowhere. On October 30, product tracker Tibor Blaho spotted a line on ChatGPT's checkout page warning that starting in December, plan usage would share a single ceiling across all ChatGPT and Codex features. Codex chief Thibault Sottiaux had said a day earlier that the team was working on the change but had not begun rolling it out.
The 28-day campaign produced a long list of improvements alongside the resets. Default speeds for GPT-6 Astra and GPT-6.1 Sol in subscription products rose about 50%. GPT-6.1 Sol's default context expanded from 272,000 to 500,000 tokens. Caching and request-routing work effectively stretched the same allowance further. GPT-6.1 Terra returned with markedly better writing and persona capabilities while consuming half the usage of 6.1 Sol. Resets, both global and banked, numbered more than ten across the cycle. On day 25 came the Pro and Business Premium doubling, accompanied by Sottiaux's line that the team was graduating "from The Reset Company to The Token Company."
New Device Limits Close a Workaround
A second restriction landed in the same window. A new help page on payment cards and device limits caps how many paid accounts a single device can use, covering Go, Plus, and the Pro 100, 200, and 500 tiers. Exceeding the cap can trigger a "Device limit reached" message at login or a mid-session logout, even after payment clears. Signing out of another account does not free a slot.
Business, Enterprise, Edu, and free accounts fall outside the limit. The rule arrives just as Codex added multi-account switching on mobile, closing off a workaround heavy users had relied on. The company points to credits purchased through the Codex app's usage and billing section as the sanctioned way to add capacity. For developers and researchers who need flexible access to older models, platforms like z2u.com offer ways to Buy OpenAI GPT-3 Access without committing to higher-tier subscriptions.
The device-limit documentation adds details that matter for anyone already holding several subscriptions. Purchases made before the restrictions took effect can continue under an existing-subscription exception, but canceling one does not guarantee a replacement will be approved. The limits attach to the payment card as well as the device, so a shared household card or a public device can trip them even when each person uses their own account. Buying credits or accepting a reset does not clear a card or device restriction.
Token Accounting Makes Compute Costs Visible
The arithmetic behind the shift is straightforward. Counting messages hides the true cost of a query: a weather question and a half-hour reasoning run looked identical on the meter, with OpenAI absorbing the difference. Token accounting turns a subscription into something closer to prepaid API capacity, with overage covered by credits and compute costs made visible to the user.
That pressure is real. After GPT-6 Astra launched in September, demand spiked hard enough that OpenAI paused new subscriptions to the $200 Pro tier. When sales reopened, Sottiaux acknowledged the new arrangement worked out to roughly half the API-equivalent spend of the old one. The frequent resets of the past month suggest the allowance is a valve the company can turn at will; now the method of adjustment shifts from handing out extra resets to uniform metering.
Competitive Pressure and Financial Stakes
The competitive backdrop sharpens the stakes. Anthropic more than doubled quarterly revenue to $11.6 billion in the second quarter from the prior period, while OpenAI's revenue rose about 18% to $6.7 billion, according to the Wall Street Journal. Anthropic posted a small operating profit for the quarter, in contrast to widening losses at OpenAI, which ran a $12.3 billion operating loss on that revenue and has filed paperwork for a listing expected next month.
For investors, the pooling change is a margin story as much as a product one. Moving to token metering shifts compute cost onto users who consume the most, a lever that matters when the company is preparing its own potential IPO as early as next year. OpenAI has secured roughly $11 billion in commitments for its latest round, against an initial $30 billion target that the company has since scaled back to as much as $20 billion. The round values it at $1.25 trillion, below the $1.4 trillion it sought, and the company plans to approach family offices and wealthy individuals for the remainder.
What Subscribers Should Know
For subscribers, the practical question is how large each tier's pool will be. OpenAI has not disclosed token allotments or the exact December start date, leaving users unable to calculate whether a combined pool will stretch further than the separate allowances it replaces. Light users who lean on Instant will likely notice little. Heavy users who spend the day reasoning in chat and the night running long coding jobs will now compete with themselves for one balance.
The era of effectively unlimited questioning narrows to a single mode. Instant chat remains free of the shared pool, but everything else now flows through one balance. For those exploring older model access alongside modern subscriptions, z2u.com provides a marketplace for GPT-3 Accounts and other OpenAI offerings, giving users flexible entry points as the pricing landscape shifts. Whether the combined pool will stretch further than the separate allowances it replaces remains the open question—one OpenAI has so far declined to answer with concrete numbers.