OpenAI faces slowdowns and errors in AI services due to Cloudflare infrastructure breakdown today (18)

Users of OpenAI’s artificial intelligence services, such as ChatGPT, face access difficulties and slow operations this Monday, November 18, 2025. The interruption occurs due to a failure in the infrastructure of Cloudflare, the network provider that supports the platform. The problem began around 11:30 am (Brasília time) and affects millions of global accesses.

US-based OpenAI confirmed the investigation into the incident on its status page. Tools like DALL-E and the Language Models API also report high errors. The company estimates partial recovery in a few hours, but did not specify exact causes.

Reports indicate that 91% of complaints involve ChatGPT, with spikes in reports above the usual volume. Platforms like Downdetector registered more than 1,000 notifications in just a few hours.

• Main services affected: ChatGPT, OpenAI API and DALL-E.
• Start time: 6:30am (Brasília time).
• Scale: Global, with a focus on regions in North America and Europe.

Services impacted by failure

Slowness directly affects con processingon ChatGPT, OpenAI’s main product. Users report 500 error messages and excessive response times.

Other tools, such as the DALL-E imager, face similar delays. The API, used by developers, records failures in integrations with external applications.

Recent outage history

OpenAI records seven critical incidents in the last 30 days, with uptime of 95.24%. On November 7, a breakdown on the Sora lasted hours.

At the beginning of August, latency errors reached 99.27% of monitored services. These events highlight dependency on providers like Cloudflare.

Reports from November 18 show 18 complaints resolved within seven hours. The company adjusts metrics to avoid recurrences.

Leia Tambem

Processador AMD Ryzen AI 7 445 registra desempenho inferior ao modelo Ryzen AI 5 340 em testes

Hatch elétrico BYD Dolphin Mini 2026 ganha ajuste de suspensão e lidera vendas no Brasil

Aparelho premium Vivo Pad 6 Pro estreia no mercado com processador Snapdragon 8 Elite Gen 5

OpenAI actions during the crash

OpenAI engineers implement mitigations from the beginning of the failure. Updates indicate recovery in the API, with a focus on conversations.

Users receive email alerts about progress. The platform recommends periodic checks on official status.

• Ongoing measures: Reduction of errors by 50% in the last two hours.
• Estimated duration: Up to four total hours.
• Support: Dedicated channels for priority reports.

Infra dependencyexternal structure

Cloudflare, responsible for data traffic, faces 500 errors in dashboards and APIs. This impacts not only OpenAI, but sites like X and Canva.

Experts point out that breakdowns in global networks expose vulnerabilities in AI ecosystems. OpenAI diversifies providers to mitigate future risks.

Resolution and continuous monitoring

The recovery is progressing, with 70% of services operational at 2pm (Brasília time). OpenAI publishes post-incident reports for transparency.

Users monitor status in real time to resume activities. Alternative AI tools win searches during these events.

The outage reinforces the need for backups in cloud-dependent operations. The company promises regular updates.