ChatGPT Down: OpenAI Restores Service After Over 12,000 Reports About Elevated Error Rates

You're not alone if ChatGPT is not functioning a while ago.

ChatGPT was down a few hours ago, and it affected people who depend on it for a living. The partial outage reached DownDetector, peaking at over 12,000 complaints.

During the downtime, several users struggled to load conversations, receive responses, or access OpenAI's platform services. This even became more alarming for professionals and developers who rely on the AI chatbot for daily workflows.

OpenAI Confirms Elevated Error Rates

ChatGPT
Discover five chatgpt apps that fall short on privacy, reliability, or usefulness, and learn when it’s smarter to stick with native tools instead for security.

According to Engadget, OpenAI quickly acknowledged the issue via its official status page, citing "elevated error rates" affecting both ChatGPT and Platform users.

Engineers deployed fixes promptly, and by 5:14 p.m. ET, OpenAI confirmed that the primary issue had been resolved. Most users regained access shortly afterward, although minor instability persisted briefly in some regions during recovery.

API Fine-Tuning Issue Still Under Watch

While core services are now operational, OpenAI noted that one component remains under active monitoring. The fine-tuning feature within the API continues to experience irregular behavior.

According to the company, mitigation steps have already been applied, and performance is steadily improving as teams closely monitor system stability.

ChatGPT outages are becoming common on the platform. The AI chatbot was hit by downtime on January 23, 2025, when thousands of users reported that ChatGPT was inaccessible.

The following month, another outage hit ChatGPT. Some users said that prompts were not working at that time.

Claude AI Faces a Similar Outage

OpenAI wasn't the only AI provider affected today. Anthropic's chatbot, Claude, experienced a separate but comparable outage earlier in the day. Users reported widespread API errors across all Claude models, disrupting developers and enterprise users.

Anthropic confirmed the problem and stated it was fully resolved by 1:00 p.m. ET.

Originally published on Tech Times