DeepSeek and Moonshot AI reportedly used OpenAI's GPT-6 Astra to train newer artificial-intelligence models, according to a report by The Information cited in coverage published on 29 October 2026.
The available reporting does not provide details about which DeepSeek or Moonshot models were involved, how the outputs were obtained, or the scale of the reported training. It also does not establish whether OpenAI has made a formal complaint or whether any regulator has opened an investigation.
Why the report matters
Using one model's outputs to help develop another is commonly discussed as model distillation. In that process, a newer system can learn patterns from generated responses rather than relying only on conventional training data. The report, however, does not explain the specific method allegedly used by DeepSeek or Moonshot.
The claim adds to wider debate over competition in artificial intelligence. OpenAI's GPT-6 Astra is identified in the reports as the model whose outputs were allegedly used, while DeepSeek and Moonshot are identified as the companies that reportedly trained newer systems with those outputs.
Coverage from Reuters, TradingView, Finimize and Crypto Briefing published on 29 October 2026 also described the development as a report that DeepSeek and Moonshot used GPT-6 Astra outputs. Those reports, as provided here, do not add verified technical details about the alleged training process.
What remains unclear
The available material does not state whether the companies used publicly accessible outputs, obtained access through an account, or relied on another method. It also does not include responses from OpenAI, DeepSeek or Moonshot.
Those unanswered questions are important because the circumstances could affect how the reported conduct is assessed. The supplied reports do not establish whether the alleged use breached a contract, violated a policy or broke any law.
For readers and businesses following AI development, the immediate significance is that model outputs may be part of competition between major AI laboratories. The available information does not support a conclusion about the performance, release status or commercial impact of any model reportedly trained in this way.
Conclusion
The Information reported that DeepSeek and Moonshot used OpenAI's GPT-6 Astra to train models, but enough detail to independently verify the method, scope or potential consequences is not yet available.
Frequently Asked Questions
Q. Which companies reportedly used GPT-6 Astra outputs?
DeepSeek and Moonshot AI were identified in the reporting.
Q. Which OpenAI model was reportedly involved?
The reports named OpenAI's GPT-6 Astra.
Q. Who first reported the claim?
The development was attributed to The Information in the supplied coverage.
Q. When was the development reported?
The related coverage was published on 29 October 2026.
Q. Is the reported training method confirmed?
The supplied material presents it as a report and does not provide enough technical detail to independently verify the method.
