OpenAI's Codex coding service has experienced repeated reliability challenges in recent weeks, including a full 56-minute outage on September 25 affecting web, API, CLI, and VS Code extension components, followed by elevated errors on September 29 and degraded performance incidents on October 7. These events, tied to internal issues amid rapid scaling, high user volumes exceeding 40 million active accounts, and frequent model updates, have driven the 80.5% market-implied probability of another outage before year-end. Traders view the pattern of mitigations, usage resets, and partial degradations as evidence of ongoing infrastructure strain rather than isolated incidents, with potential catalysts including further model launches or load spikes that could trigger additional disruptions.