Leaks point to unreleased Claude and GPT models in testing
Decrypt reports that users say Claude is routing some requests to a stronger unreleased model, that a Mythos 5.5 label appeared in Anthropic's site code, and that OpenAI appears to be testing GPT-6.2 Astra. The report attributes these indications to leaks and user reports, with no confirmation stated.
Why “Unverified”
- Reported by Decrypt alone, with no second source or primary confirmation yet
1 independent source group across 1 post · evidence score 56/100. How labels work
What the sources claim
-
Users report that Claude is sending some requests to a stronger unreleased model.
“Users say Claude is routing some requests to a stronger unreleased model”
Decrypt -
A Mythos 5.5 label appeared in Anthropic's site code.
“a Mythos 5.5 label appeared in Anthropic’s site code”
Decrypt -
OpenAI appears to be testing a model called GPT-6.2 Astra.
“OpenAI appears to be testing GPT-6.2 Astra”
Decrypt
Sources
More in AI & Tech
Summary and claims extracted by AI with prompt analyze-story@5 on 2026-11-02 23:20 UTC. Labels are computed from the sources, not by the model.