Is OpenAI preparing a $500-per-month Pro Max plan?

What exactly was found in OpenAI's configurations

I would not call Pro Max a finished product yet: as of September 26, 2026, OpenAI has not published an official announcement. However, internal interface configurations revealed a very specific price: $500 per month before taxes, plus a $600 option including VAT.

That is 2.5 times the price of the current $200 Pro plan. The discovered description says “Fastest Work + Fastest Codex,” suggesting that the tier is selling more than extra requests: it is primarily selling faster execution for work and coding tasks.

There is also a confirmed part of the story. OpenAI's Help Center states that new subscriptions and upgrades to the $200 Pro plan have been temporarily paused since September 10, while existing subscribers retain access. It does not state why Pro Max may be appearing.

• Price: $500 per month before taxes at the time of the leak.
• Positioning: maximum speed for Work and Codex.
• Timing: the configuration surfaced five days before DevDay, scheduled for September 29.

The theory that Cerebras hardware is involved and could deliver a tenfold speed increase remains speculative. OpenAI has not confirmed it, so it is premature to tie a future tier to specific chips. The configuration label reveals a product direction, but says nothing about consistent latency, actual limits or capacity availability under load.

Why selling speed changes the subscription model

If Pro Max does launch, OpenAI will more clearly separate users not only by model access, but also by compute priority. That makes a tangible difference for agentic workflows: a long chain of calls turns the latency of every step into latency for the entire task.

Codex users and people running demanding work processes stand to gain most, especially where execution time has measurable value. The ordinary subscriber loses out if fast access becomes a separate, expensive resource while the former Pro tier remains restricted or closed to new sign-ups.

My first engineering question is not about peak speed, but about predictability. Without published limits, prioritization rules and behavior during busy periods, “Fastest” is a compelling label rather than a technical guarantee. The real news will come when OpenAI explains exactly what the extra $300 per month buys.

We previously examined how AI agent marketplaces are turning autonomous agents into deployable business tools, while introducing new integration and security trade-offs. That context helps frame what a premium OpenAI model could mean for agent capabilities and adoption.