OpenAI Opens 1 Million Token Context Access in Codex for ChatGPT Users

OpenAI activated a feature that had remained restricted. Codex, the company's AI-assisted programming tool, now allows ChatGPT Plus and Pro subscribers to access the extended context window of the GPT-5.6 Sol model, supporting approximately 1.05 million tokens. This capability was previously available only through API keys. The announcement came from Tibo, an engineer responsible for the platform, via X. The feature was disabled by default but is now available to subscribers, allowing retention of larger volumes of code, logs, and conversation histories before automatic compression occurs. The expanded capacity proves useful for scenarios involving large repository refactoring, complex debugging, and extended sessions. The company emphasized that Codex's default configuration was optimized by balancing performance and operational cost. Tibo noted that expanding the context remains a user choice, though he stressed that the defaults were carefully tuned. Users reported that activating maximum context significantly accelerates monthly token consumption, potentially exhausting allocations during intensive use. Developers suggested the feature be activated only when essential, preserving day-to-day economic efficiency.