DeepSeek and Moonshot AI reportedly relied heavily on GPT-6 Astra outputs

The Information reports that DeepSeek and Moonshot AI relied heavily on outputs from OpenAI's GPT-6 Astra to train their latest models. Reuters attributed the finding to The Information, which it said cited people with direct knowledge of the matter.

Reuters also reported that, according to The Information, Moonshot's Kimi K3.1 Pro and DeepSeek's upgraded V4.1-Flash 1017 benefited from distilling Astra. Reuters described Kimi as released the preceding week and the DeepSeek upgrade as launched Oct. 17.

The Information placed the reported shift alongside Anthropic's effort to stop Chinese labs extracting capabilities from its Claude models. Anthropic said its safeguards reduced data harvested in distillation attacks by about 90% to 99%, Reuters reported. The reporting does not establish that Anthropic's measures drove either lab to Astra. OpenAI, DeepSeek and Moonshot AI did not immediately respond to Reuters' requests for comment at the time of its report.