OpenAI platform users report sightings of GPT-6.2 Astra model
Evidence of a 'GPT-6.2 Astra' model has surfaced in OpenAI's Codex usage statistics, suggesting the company is conducting early-stage testing of a new iteration.
Users of OpenAI's Codex platform have reported the appearance of a new model identifier, 'gpt-6.2-astra,' within their personal usage statistics. This development suggests that the company may be running silent tests or A/B deployments of an unannounced successor to the GPT-6 Astra model, released in September 2026.
OpenAI has not provided official confirmation, nor has the model appeared in public API directories or changelogs. According to reports from TestingCatalog, some accounts show a significant percentage of requests directed toward this new identifier, particularly in late October data. This pattern is consistent with common industry practices for internal or phased software testing.
The appearance of the new model aligns with similar reports surrounding competitors, such as Anthropic, where users have noted hidden references to potential new models like Fable 5.5 and Mythos 5.5. This activity highlights the current competitive climate for LLM developers as they push for iterative performance upgrades.
Beyond technical capabilities, users are closely monitoring the impact on token consumption and account limits. With recent policy changes linking ChatGPT and Codex usage quotas, the cost-efficiency of potential future models like GPT-6.2 Astra remains a primary concern for the platform’s user base.