OpenAI, the creators of ChatGPT, recently made a notable shift in their approach after users expressed frustration over the removal of older models, which left only the new GPT-5 available for use.
What’s the story? The anticipation surrounding the new GPT model reached a peak with a livestream announcement on August 6.
- On August 6, Sam Altman, the CEO of OpenAI, introduced the newest model designed to enhance ChatGPT – GPT-5.
- Following the launch, OpenAI eliminated access to older models, compelling all users to switch to the latest version.
- In a surprising turn, OpenAI has now allowed ChatGPT Plus subscribers (paying $20/month) to access legacy models, though only the model 4o is currently available.
What sparked the change? Users had formed strong connections with the responses generated by GPT-4o, customizing prompts that elicited preferred outcomes.
- Previously, ChatGPT offered multiple models, with versions o3 and 4o catering to complex tasks like advanced reasoning and coding.
- With GPT-5 expected to merge the best features of its predecessors, OpenAI opted to remove legacy models to streamline the user experience.
- User backlash was swift, with Reddit inundated with complaints. One user even reported feeling nauseated at the thought of losing previous models, as many perceived GPT-5 as overly sanitized.
- During a recent Reddit AMA, users voiced their dissatisfaction, with one remarking that GPT-5 felt like “wearing the skin of my dead friend,” reflecting their attachment to GPT-4o.
- Although Altman initially mentioned that older models might return, he later confirmed that the features would be accessible to all users.
Why is this significant? OpenAI appears to have lost a number of subscribers discontented with GPT-5, although this decline may not be significant. Nevertheless, the company is focused on regaining the trust of its user base, as seen in their prompt response to feedback.
The contrasting opinions
- Many users have praised GPT-5 for its improved practicality, including better multitasking and coding capabilities.
- However, some have criticized its writing abilities when compared to GPT-4o.
- OpenAI aims for GPT-5 to serve as a more comprehensive tool rather than just a companion. Altman acknowledged in a recent tweet that the company “underestimated how much some of the things that people like in GPT-4o matter to them.”
- This new model is designed to hallucinate less frequently and adopt a more professional tone, balancing providing safe responses while addressing legitimate concerns.
What’s next for OpenAI? Altman and his team are clearly adaptable when it comes to refining the model. OpenAI now allows Premium users to access 3,000 deep reasoning queries weekly, a significant increase from previous limits.
- During the AMA, Altman inquired whether users preferred to stick with model 4o exclusively or if a new GPT-4.5 model was necessary.
- He has confirmed that although the platform is still somewhat unstable during the rollout, stability has been achieved for Plus users, who spend $200/month, whereas lower-tier users may still experience issues.
The rollout of GPT-5 has certainly had its challenges. There were various unsettling announcements that raised eyebrows among AI experts. If you are among the users, continue to engage with the various models, and do share your experiences regarding any notable differences you observe.
What makes GPT-5 different from earlier versions?
GPT-5 is designed to be a more versatile tool, integrating features from earlier models, while aiming for precision and professionalism in its outputs.
Why did OpenAI decide to remove the older models?
OpenAI aimed to streamline the user experience by removing legacy models, promoting GPT-5 as the singular choice for all users.
How are users reacting to the changes brought by GPT-5?
User reactions have been mixed, with many expressing disappointment over the perceived loss of personality and depth compared to older models.
What are the advantages of using GPT-5 over previous models?
GPT-5 boasts enhanced practical capabilities, allowing for improved multitasking and coding, though some users find its writing less engaging than previous iterations.
What future changes can we expect from OpenAI?
OpenAI seems open to user feedback, suggesting potential tweaks and new model introductions based on user preferences and requirements.
Looking for more insights or updates on the latest in AI? Don’t hesitate to explore related content at Moyens I/O.