OpenAI Re-implements Usage Caps for Plus and Business Standard Tiers

Reports are surfacing via Tell HN that OpenAI has decided to bring back restrictive usage limits for its paid subscribers. Specifically, users on the Plus and Business Standard plans are finding that their access to high-end models is no longer as expansive as it recently seemed.

The Core Issue: Return of the Cap

For a period, many users felt the constraints on GPT-4o had loosened, but the "5-hour limit" (or similar restrictive windows) has returned. This means that after a certain volume of prompts, users are throttled or forced to switch to a less capable model.

Unlimited high-speed access $\rightarrow$ Strict quota enforcement.

The Mathematical Perspective

The limit can be viewed as a function of messages ($m$) over a specific time window ($t$):

$\text{Usage Status} = \begin{cases} \text{Active} & \text{if } m < \text{Limit} \\ \text{Throttled} & \text{if } m \geq \text{Limit} \end{cases}$

Where the time window $t$ is typically measured in 3 to 5-hour increments.

---

Comparison of Impact

The following table outlines how different tiers are currently experiencing these changes:

---

User Experience Flow

The process of hitting the limit generally follows this logic:

Community Reaction

Users are expressing frustration over the lack of transparency regarding these changes. Many feel that the "Plus" experience is becoming less predictable.

"I was in the middle of a complex coding project and suddenly got hit with the limit. It's incredibly disruptive to the workflow when the goalposts keep moving." — Anonymous HN User

Technical Simulation

If the API were to return a status for this limit, it might look something like this:

{
  "error": {
    "message": "Rate limit reached for model gpt-4o",
    "type": "insufficient_quota",
    "param": null,
    "code": "rate_limit_exceeded"
  },
  "reset_time": "2023-10-27T14:00:00Z"
}

---

Summary Checklist for Users

If you suspect you have been throttled, you can check the following:

• Check if you are using GPT-4o or GPT-4o-mini.
• Look for the "You've reached your limit" notification in the chat UI.
• Verify if the issue persists across different devices.
• Monitor the reset timer provided by the interface.

In conclusion, while OpenAI continues to iterate on its models, the infrastructure costs associated with GPT-4o seem to be driving a return to more rigid usage policies for the mid-tier paid users.