OpenAI GPT-5 Issues: Double Rate Limits for Paid Users
GPT-5: Initial Impressions, Issues, and OpenAI’s Rapid Response
Table of Contents
GPT-5 has arrived, and the initial response is…complex. While boasting impressive coding speeds, OpenAI’s latest large language model (LLM) isn’t without its quirks. From struggles with creativity to unexpected limitations in output length, early users have encountered a few bumps in the road. But don’t worry, OpenAI is already on it, rolling out updates and addressing concerns as quickly as they arise.Let’s dive into what we know about GPT-5 so far, the issues users are facing, and what OpenAI is doing to fix them.
GPT-5: A Coding Powerhouse, But Not a Creative Genius
Early tests, including those conducted by BleepingComputer, reveal that GPT-5 excels in coding tasks. It substantially outperforms previous OpenAI models, like o3, in speed and efficiency.If you’re a developer,this is a major win – GPT-5 could dramatically accelerate your workflow.
However, the picture isn’t entirely rosy. Users are reporting that GPT-5 struggles with tasks requiring creativity in writing. It can feel…flat. More concerningly, the model isn’t always activating its advanced reasoning capabilities when prompted, leaving users feeling like they’re interacting with a less intelligent version of what was promised.
And then there’s the length issue. Manny users have noticed that GPT-5 frequently produces shorter responses than expected, even when explicitly asked for more detail. This can be frustrating if you need comprehensive answers or in-depth content.
The Throttling Debate and Sam Altman’s Description
The short output length sparked speculation that OpenAI might be intentionally throttling token output to reduce costs. Tokens are essentially the building blocks of text that LLMs process, and limiting them can save money. However, OpenAI CEO Sam Altman quickly dismissed this theory.
In a post on X (formerly Twitter), Altman explained that a bug in the “autoswitcher” – the system responsible for activating GPT-5’s advanced features – was to blame. “[Y]esterday, the autoswitcher broke and was out of commission for a chunk of the day, and the result was GPT-5 seemed way dumber,” he wrote.
This explanation offers a reassuring perspective, suggesting the issues aren’t intentional limitations but rather technical glitches that are being actively addressed.
openai’s Immediate response: Rate Limits, Reasoning Toggle, and Model openness
OpenAI isn’t sitting still.They’re responding to user feedback with a series of updates designed to improve the GPT-5 experience. Here’s what’s happening:
Increased Rate Limits: OpenAI will double GPT-5 rate limits for GPT Plus subscribers.This means you’ll be able to send more requests to the model within a given timeframe. However, it will take a few days for these changes to roll out.
4o Model Option: You’ll soon be able to continue using the 4o model, even with GPT-5 available. OpenAI will monitor usage to determine how long to support legacy models.
Reasoning Toggle: A new toggle is being tested that will force GPT-5 to utilize its reasoning capabilities. This should address the issue of the model not activating its advanced features when expected.
Improved Transparency: OpenAI plans to make it clearer which model is answering your queries and will update the user interface to make it easier to manually trigger the “thinking” process. This added control will empower you to get the moast out of GPT-5.
GPT-5 Rollout and Access to Legacy Models
The rollout of GPT-5 is still in progress,and access is being granted to both paid and free users gradually. It may take some time before everyone has access to the new model.
If you’re a GPT Plus customer and want to access older models like 4o,you’ll currently need to subscribe to the $200/month Pro plan. OpenAI is planning to restore access to these legacy models for Plus customers in the future.