Chinese AI laboratories mainly turn to GPT-6 Astra, the most powerful OpenAI model, instead of Claude from Anthropic. The Decoder reports this in the metadata of a publication from 29 October 2026 year.
For the market, this could mean a shift in lab preferences toward OpenAI's model. However, the available data do not show how many organizations are affected, what access methods are used, or what exactly the strengthened defenses of Anthropic imply.
editorial commentary
Why it matters
Likely consequence — increased attention to GPT-6 Astra by Chinese AI laboratories. The next observed signal will be independent confirmation of such a shift or the appearance of data about specific labs and reasons for the choice. Significant uncertainty is due to the conclusion being based on one metadata recap without a primary statement.