OpenAI Reopens $200 ChatGPT Pro Sign-Ups — But Cuts API Credits in Half in Shift Toward Pay-Per-Use

ChatGPT Pro's $200 tier returns to new sign-ups with the 5-hour cap gone and GPT-6 Sol and Luna at half price — but API credits per dollar are halved.

OpenAI is reopening its $200-per-month ChatGPT Pro subscription to new sign-ups, ending a pause that began after demand for its newest models overwhelmed capacity. But there is a catch that has not gone unnoticed: the company is simultaneously cutting the API credits Pro subscribers receive per dollar in half, a change that has sparked controversy among power users.

The news was shared by OpenAI employee Thibault Sottiaux, who said the plan would reopen, and was detailed in coverage by The Decoder and other outlets on September 29. The re-opening follows OpenAI's decision weeks ago to pause Pro sign-ups as demand for GPT-6 Astra exceeded available capacity — and it arrives with a noticeably restructured deal. For continuous coverage of OpenAI's pricing moves and everything else in AI, follow the latest AI news on AI Buzz Wire.

What Changes for Pro Subscribers

The revamped Pro plan mixes good news with bad:

The 5-hour usage cap is gone for good. Subscribers can now spread their weekly allotment however they want, rather than racing against a rolling short-window limit.

API credits per dollar are cut in half. The subscription's effective usage value drops substantially, pushing heavy users toward paying for API usage directly.

Better underlying models offset the cut, OpenAI argues. Sottiaux pointed to GPT-6 Sol and GPT-6 Luna, which shipped this week at half the API price of their predecessors, claiming subscribers should still get more done than a month ago.

That last point is the crux of the debate. If a task costs half as much to run but you get half the credits, your effective output is unchanged — the value gain only materializes for subscribers whose workloads benefit disproportionately from the newer, cheaper models. Users whose usage does not map cleanly onto the new models see their subscriptions quietly shrink.

The Strategic Shift: Away From Subsidized Subscriptions

The restructure signals something bigger than a pricing tweak. OpenAI is moving away from heavily subsidized flat-rate subscriptions toward pay-per-use pricing — a model Microsoft has already applied to Copilot in some areas.

Sottiaux was explicit about the endgame, writing that over time, API prices should fall far enough "that it makes sense for most to buy usage as needed without there being a significant gap between what you get in a subscription and what you get in the API for a dollar spent."

The economics make sense for the lab. Flat-rate plans are a gamble: the heaviest users extract the most value while paying the same as casual ones, and computing those losses is exactly what a $200 tier with generous caps does. Usage-based pricing ties revenue more tightly to the actual value customers pull from AI — and it insulates the company from the usage asymmetry problem that has plagued every unlimited tier in the software industry.

Why OpenAI Is Doing This Now

The timing is not random. Demand for frontier models has repeatedly slammed into OpenAI's capacity limits this quarter, from the Astra-driven Pro pause to rate limits across API tiers. When capacity is the binding constraint, underpriced unlimited plans actively work against the business: every subsidized heavy user crowds out a full-price API customer.

Halving credit value while reopening the plan is, in effect, a capacity-management decision dressed as a pricing update. It throttles the most intensive subscription workloads while keeping the headline product available — and it converts the most demanding users into metered API customers, where their usage is finally priced at something close to cost.

The reopening also resolves an awkward limbo for prospective subscribers. Since the pause, anyone who wanted the $200 tier had no option but to wait, downgrade to the Plus plan, or run their workloads through the metered API. Restoring access — even on tighter terms — gives OpenAI's most demanding retail users a formal home again, while the restructured credits quietly reset expectations for what a flat monthly fee buys.

What It Means for Subscribers and the Market

For subscribers, the practical advice is to audit usage. Workloads that run on GPT-6 Sol and Luna will stretch roughly as far as before, with better scheduling flexibility now that the 5-hour cap is gone. Workloads that depend on older premium models, or that spike in short bursts, will hit the halved credits sooner.

For the market, the move is another data point in the industry's pricing consolidation. Rival labs face the same usage-asymmetry math on their own premium tiers, and providers across the industry are converging on the same conclusion — subscriptions for casual and prosumer use, metered billing for serious workloads.

The open question is churn. Power users who feel the halving most are precisely the vocal community whose advocacy fuels a product like ChatGPT Pro. If OpenAI's bet pays off, usage-based pricing becomes the norm across consumer AI; if it drives the heaviest users to rival tools with softer caps, the halved credits may yet be walked back.