OpenAI's Codex Coding Tool Reportedly Down for Some Users as Outage Reports Spike Overnight

OpenAI's Codex, the company's AI-powered coding assistant, was reportedly experiencing connectivity and access issues for some users early Tuesday, with outage tracking services and user reports pointing to intermittent disruptions affecting the tool's desktop application and related services.

According to outage tracking site StatusGator, 21 user-submitted reports of problems with Codex were logged over a 24-hour period ending Monday evening, with the platform describing OpenAI as experiencing a partial outage at the time. The social media account Status Is Down flagged the issue early Tuesday morning, asking followers whether they were affected and using the hashtags #OpenAI, #OpenAIDown, #Codex and #CodexDown as reports of the disruption circulated online.

Multiple users also reported problems directly on OpenAI's GitHub issue tracker for the Codex project throughout the day Monday and into Tuesday. Several issues logged by developers described connectivity problems affecting the Codex desktop application, including disconnection errors related to networking and endpoint connectivity, as well as separate issues involving the tool's integration with Model Context Protocol servers, a framework Codex uses to connect with external tools and data sources. Additional reports cited problems with tool-calling functionality and issues specific to the Codex application running on Windows systems.

OpenAI's official status page acknowledged ongoing issues tied to its systems, though the company's most recent public update focused specifically on disruptions affecting FedRAMP workspaces, a designation for cloud environments that meet federal government security compliance standards. According to that update, OpenAI said core functionality had been restored following an earlier disruption, but noted continuing issues affecting Codex, workspace analytics, conversation search, the ability to search for custom GPTs, ChatGPT user invites, and the Compliance Logs Platform download feature specifically within FedRAMP-designated environments. The company said it was continuing to work on resolving those remaining issues and would provide further updates as more information became available.

The disruption adds to what has been a bumpy stretch for Codex's reliability in recent weeks. According to outage history compiled by StatusGator, OpenAI has logged a series of warning-level service disruptions on nearly a daily basis over the past several weeks, including periods of degraded service lasting anywhere from several hours to a full day at a time between late June and early July. One previously resolved incident specifically affecting Codex and related FedRAMP services lasted approximately five hours and 48 minutes, beginning in the early morning hours of July 1.

Codex has also experienced a range of other technical issues in recent months unrelated to Tuesday's reported disruption. According to incident logs maintained by monitoring service IncidentHub, Codex has dealt with several previously resolved issues over the past few weeks, including a period in late June when usage limits within Codex appeared to deplete faster than expected, along with earlier incidents involving access token errors, a "Selected Model is at Capacity" error message, elevated error rates specifically affecting the GPT 5.5 model within Codex, and increased latency during a process the company refers to as Codex compaction.

OpenAI has periodically addressed broader Codex-related errors through its own developer community forum in the past, acknowledging increases in error rates and confirming that engineering teams were actively working to resolve underlying issues as quickly as possible. The company has generally directed users to its official status page for real-time updates during periods of degraded service, a practice that appeared to continue during Tuesday's reported disruption.

Codex, first introduced by OpenAI as an AI coding tool designed to help developers write, debug and manage code through natural language prompts, has become an increasingly central part of the company's broader product lineup as demand for AI-assisted software development tools has grown across the technology industry. The tool is available both through a web interface and as a downloadable desktop application, with additional integrations available through command-line interfaces used by many professional software developers.

Outage tracking services generally caution that reported issue counts reflect self-submitted user data and public signals such as social media activity, meaning the true scope of any given disruption can sometimes differ from the volume of individual reports logged at a particular time. Even so, the combination of user reports on GitHub, social media posts flagging the outage, and OpenAI's own acknowledgment of ongoing technical issues within certain workspace environments suggested that at least some portion of Codex's user base was experiencing meaningful disruption to the service as of early Tuesday.

For affected users, common troubleshooting steps recommended by technical support resources typically include checking OpenAI's official status page for the most current information, verifying internet connectivity, restarting the Codex application, and, where relevant, checking whether an issue is isolated to a specific integration such as an MCP server connection rather than reflecting a broader outage of the underlying service. Developers experiencing persistent issues are generally encouraged to file detailed reports through OpenAI's GitHub repository, where the company's engineering team can track and triage individual bug reports alongside broader service-wide disruptions.

As of Tuesday morning, OpenAI had not issued a comprehensive public statement addressing the full scope, cause or expected resolution timeline for the reported Codex disruptions occurring outside the FedRAMP environment specifically referenced in its most recent status update. The company's history of frequent, relatively short-duration service warnings over the preceding weeks suggests that intermittent disruptions of this kind have become a recurring, if generally short-lived, feature of Codex's operation as OpenAI continues to scale the tool's usage among a growing base of developers relying on it for day-to-day coding tasks.

Users experiencing ongoing problems with Codex are encouraged to monitor OpenAI's official status page at status.openai.com for the most current information, as the company works through what appears to be a mix of both newly reported connectivity issues and previously acknowledged, more narrowly scoped disruptions affecting specific workspace configurations. Given the pattern of past incidents, similar disruptions affecting Codex have historically been resolved within a period ranging from several hours to roughly a day, though OpenAI has not provided a specific timeline for full resolution of Tuesday's reported issues as of this report.