DeepSeek will compensate users after a major service disruption took its chatbot and application programming interface services offline for nearly 5 hours on Thursday. The Chinese artificial intelligence company said it would provide credits covering eligible consumption between October 8 and October 14.

The outage affected DeepSeek's namesake chatbot and API services linked to its V4 Pro and V4.1 Flash models. Services were restored after 4 hours and 50 minutes, with the company reporting recovery at around afternoon 1:00.

The disruption was felt by users globally. Some ed error messages on social media, including notices that the server was temporarily unavailable.

Refund and service review

DeepSeek's compensation will take the form of credits rather than a cash payment, according to a notice on the start-up's API platform. The company did not provide further details about how the credits would be applied in the supplied notice.

DeepSeek said it had begun a broad investigation and corrective work focused on service stability. It also said the review had addressed single points of dependency, describing the changes as an effort to prevent similar disruptions from happening again.

The company apologised for the interruption and the disruption experienced by users. It did not explain the specific cause of Thursday's outage.

Pressure from rising demand

The incident comes as DeepSeek faces growing pressure on its infrastructure while demand for computing resources increases. The company is seeking to expand its teams and computing capacity as competition in the artificial intelligence sector becomes more intense.

A Bloomberg report said DeepSeek has a plan to add about 1 gigawatt of computing capacity in China's northern Inner Mongolia autonomous region. That expansion was presented as part of the company's wider effort to increase infrastructure.

DeepSeek has experienced other service interruptions. The latest incident was described as among the most serious disruptions it has faced in recent months.

In March, the platform suffered another major interruption that continued for around 12 hours and affected hundreds of millions of users. The company also faced denial-of-service attacks in late January last year, shortly after releasing its R1 reasoning model and making it free globally.

At that time, DeepSeek described the attacks as large-scale malicious attacks. Denial-of-service attacks are intended to overwhelm servers and bandwidth with large volumes of internet traffic, according to the supplied report.

Conclusion

DeepSeek has restored its chatbot and API services and offered credits for consumption between October 8 and October 14 after the nearly five-hour outage. The company says it is reviewing service dependencies, but has not disclosed the incident's specific cause.

Frequently Asked Questions

Q. What happened to DeepSeek's services?

DeepSeek's chatbot and API services went offline for nearly 5 hours on Thursday. They were restored after 4 hours and 50 minutes, at around afternoon 1:00.

Q. Which DeepSeek services were affected?

The disruption affected the namesake chatbot and API services for the V4 Pro and V4.1 Flash models.

Q. What compensation is DeepSeek offering?

The company said it would issue users credits covering consumption between October 8 and October 14.

Q. Did DeepSeek explain the outage's cause?

No. The company said it had launched an investigation and corrective work but did not elaborate on the cause.

Q. Has DeepSeek experienced earlier disruptions?

Yes. The platform had a disruption lasting around 12 hours in March and faced denial-of-service attacks in late January last year.