GPT-6.2 Astra Appears in Codex Data as Claude Mythos 5.5 Surfaces in Anthropic Website Code

On November 1, 2026, users reported seeing a GPT-6.2 Astra model tag in OpenAI’s Codex usage data, although the model selector still displayed GPT-6 Astra. According to ChainCatcher, a screenshot posted the same day showed Claude Mythos 5.5 in hidden code on Anthropic’s Mythos webpage, while the visible page identified the model as Claude Mythos 5.1. A usage chart covering October 26 to November 1 showed GPT-6 Astra at 80% and GPT-6.2 Astra at 19.8%; another account shared a purported API response with “gpt-6.2-astra” in the response.model field. Qbit reported that GPT-6.2 Astra was being tested with a limited number of users, with access largely restricted to the $500-per-month Pro plan. Some users said about 20% of their requests were routed to version 6.2, while others said all their requests went to it. The Claude Mythos 5.5 text reportedly appeared briefly in Anthropic’s website code and was removed several hours later. Anthropic has not confirmed the model, and no archived copy of the code change has surfaced beyond the screenshot. Earlier reports said Fable 5.5 appeared in Microsoft Azure and MLflow listings, with a 1 million-token context window and a 128K-token output limit. A review found that a bot added an “azure_ai/claude-fable-5-5” entry to open-source tool LiteLLM’s pricing list on October 13, 2026; it was removed the next day with a note that Azure did not list the model. The listed specifications matched Fable 5.1. OpenAI and Anthropic have not officially released the models mentioned.