Chinese AI Labs Reportedly Distilled GPT-6 Astra
The Information reports, citing people with direct knowledge, that DeepSeek and Moonshot AI heavily used GPT-6 Astra outputs to train their latest models. The reported models include Kimi K3.1 Pro and V4.1-Flash 1017, while The Decoder links the shift away from Claude to Anthropic's tighter distillation defenses. DeepSeek and Moonshot AI did not comment, and TNW said it had not independently verified the report.
The report highlights competition over model distillation and the growing use of frontier-model outputs to train competing AI systems, though the claims remain unconfirmed by the labs involved.
This source provides an excerpt. Continue reading on the original site.
#OpenAI#GPT-6 Astra#DeepSeek#Moonshot AI#Kimi K3.1 Pro#V4.1-Flash 1017#Anthropic#Claude