OpenAI Confirms ChatGPT and API High Latency Issue Has Been Resolved

Regarding the slow response of ChatGPT and API services on May 27, OpenAI has completed the repair work from late yesterday night to early today, and the relevant services have now returned to normal.

1. Event Review

Issue Confirmation: At 22:47 Beijing Time on May 27, the OpenAI official status page confirmed that there was a "high latency" failure in ChatGPT and API services. According to user feedback from multiple regions, this issue started on the same day early in the morning, characterized by significantly longer waiting times for responses after users submitted questions.
Repair Progress: After team investigation and emergency repairs, OpenAI officially confirmed on May 28 at 04:06 Beijing Time that the failure had been resolved and services were back to normal access.

2. Remaining Observations

Although the main high latency issue has been resolved, the OpenAI status page indicates that some auxiliary functions are still under optimization and observation:

Codex Function: The context compression speed is still slower than expected.
Enterprise Edition Function: There are still some abnormal situations when switching workspaces in the Android version of the ChatGPT Enterprise Edition.

OpenAI is continuously following up on the resolution of these details. Users are advised to pay attention to the latest updates on the official website's status page if they encounter any interface operation issues during use.

OpenAI Expands Access to Advertising Platform, Invites More Advertisers, Offers Precise Targeting Opportunities for Free ChatGPT Users

OpenAI accelerates its advertising commercialization, opens the platform to more advertisers, and allows targeted advertising among free ChatGPT users in the US, Canada, Australia, and New Zealand. This move transitions from a limited test to a self-service model, aiming to balance user experience with sustainable monetization.

Altman Reverses His Statement: The Impact of AI on White-Collar Jobs Is Not as Severe

OpenAI CEO Sam Altman recently admitted that earlier predictions about AI eliminating many entry-level jobs were inaccurate. At a tech conference, he stated that while technological forecasts were roughly correct, the social impact differed from expectations, with the job market not being disrupted by AI as anticipated, marking a subtle shift in his stance.....

The 'Codex Self-Distillation' Gameplay Goes Viral! OpenAI Employees Reveal Efficient and Time-Saving AI Assistant Secrets

OpenAI programmer VB shares Codex self-distillation tips: using simple prompts to enable Codex to automatically scan historical conversations, identify repetitive manual tasks, and package them into reusable tools, significantly boosting efficiency. After receiving user feedback, VB quickly released version 2.0 of the prompts for further optimization.....

OpenAI's reasoning model independently solves an 80-year-old problem: from research assistant to scientific discoverer

OpenAI's general reasoning model independently solved the Erdős unit distance conjecture, a problem that had puzzled the academic community for 80 years in the field of combinatorial geometry, without human guidance. The problem was proposed by mathematician Erdős in 1946, and its core is to explore how many pairs of unit distances can exist between n points on a plane.