DeepSeek and Moonshot trained new models on OpenAI's Astra, report says

The Information reported on October 29 that DeepSeek and Moonshot AI leaned heavily on outputs from OpenAI's GPT-6 Astra to train their latest models, citing people with direct knowledge. Moonshot's Kimi K3.1 Pro, released last week, and DeepSeek's V4.1-Flash 1017, launched October 17, both benefited, according to the report.

The claim lands a month after OpenAI said it had traced part of a distillation campaign beginning in July to people associated with Moonshot AI. Anthropic said earlier in October that its countermeasures had cut harvested data in such attacks by roughly 90% to 99%.

Distillation lets a rival reproduce advanced capabilities faster and more cheaply, the labs argue, which is why the technique has become a flashpoint as US-China AI competition intensifies. OpenAI, DeepSeek and Moonshot did not immediately respond to requests for comment.