Want smart insight into your inbox? Enterprise AI, only what matters to data and security leaders, sign up for our weekly newspapers. Subscribe now
Openai co-founder and CEO Sam Altman is publicly accepting the major hiccup of GPT -5 yesterday’s rolloutThe company’s new, major major language model (LLM) – has yet been advertised as its most powerful and capable.
Answer user questions In a reddit ama (ask me anything) thread And In a post on X today afternoon, ALTMAN accepted many issues that disrupted the launch of GPT-5, including defective model switching, poor performance and user confusion – Inspire Openai to partially to return some of its platform changes and Restore user access to earlier models like GPT-4o.
Altman wrote, “It was a bit more bumpy as we expected,” Altman wrote In Answer a question on reddit About Big GPT -5 launch.
Related to Wrong model display chart Openai’s GPT-5 Livestream shown during Altman Said, “People were working late and were very tired, and the human error went on the way. There is a lot for a livestream in the last hours. ,
AI scaling hits its boundaries
Power caps, rising token costs, and entrance delays are re -shaping Enterprise AI. Join our exclusive salons to learn about top teams:
- Transform energy into a strategic profit
- Architecting efficient estimates for real thrruput benefits
- Unlocking competitive ROI with sustainable AI system
Secure your location to stay ahead,
While he noted that the blog posts and system cards were accurate together, misunderstandings already carried forward the launch that faced investigation from initial users and developers.
Problems with new automatic model router
According to ALTMAN, an important cause of the trouble stems from Openai’s new automatic “router”, which indicates the user to one of the four GPT-5 variants-for heavy logic functions with-mini, nano and pro-one optional “thinking” mode.
On X, Ultman revealed that an important part of that system-autosquill-“was out of commission for a part of the day,” the “way of” the way “is visible due to GPT-5.
In response, Openai says that this model is implementing changes in the decision limit and will make it more transparent that the model is responding to a given query.
A UI update is also on how to help users to trigger the thinking mode manually.
Additionally, Altman confirmed that Openai will now allow Chatgpt Plus users to continue using GPT-4O-pre-default model -After a wave of complaints about inconsistent performance of GPT-5. He Said on reddit The company is “Till how long it is trying to collect more data on the tradeoff before deciding how long to offer the heritage model.
Still including many users Wharton School of Business Professor Ethan Molik like openi beta tester Confused and expressed disappointment Openai unilaterally upgrade its Chatgpt experiences in GPT-5 and initially reaches older models.
The performance of the real world is behind the promotion
Openai’s internal benchmark can show GPT-5 showing the pack of LLMSBut real -world users are sharing a different experience.
Since the launch, users have posted several examples of GPT-5 that created basic errors in mathematics, logic and coding works.
data scientist Colin Fraser has posted screenshots of GPT-5, incorrectly resolved whether 8.888 is equal to 9 (this is not clear)While another user showed it a simple algebra problem: 5.9 = x + 5.11.
And yet other users reported trouble in receiving accurate answers to mathematics term problems or using GPT-5, so that it could debug his presentation chart.
Developer feedback is not much better, with Users spoiled GPT images “a-shot” in certain programming tasks -Completing them well with a single-prompt- Compared to the new model Cloud Opus 4.1 of rival AI Lab Anthropic.
And security firm SPLX found that GPT-5 still suffers from serious weaknesses Until its safety layer is hardening, to accelerate the injection and obfusted logic attack.
Openai in Spotlight
With 700 million weekly users on Chatgpt, Openai remains the biggest player in AI generated by viewers.
But that scale has given rise to increasing pain. Altman mentioned in his X post that The API traffic exceeded 24 hours after the traffic GPT-5 launch, which contributes to platform instability.
In response, Openai says that this will make double rates for the Chatgpt Plus users, and will continue to twist the infrastructure as it collects the response.
But initial misunderstandings-UX changes in a high-profile launch and confusing errors are complicated. Opened a window for rivals to achieve land.
The pressure for Openai is to prove that GPT-5 is not just an incremental update, but a true step ahead. Depending on the initial rollout, many users are not confident – yet.
