--- format: "aidr-story-markdown/v1" id: "8ed31c0d6427dab8728ce16f47e9092d41f4530061dc33ed673af6ee3caf975d" canonical_url: "https://aidr.today/8ed31c0d?lang=en" title: "Chinese AI Developers Use OpenAI Astra to Train New Models Following Anthropic Blocks" lang: "en" requested_lang: "en" available_langs: ["en","vi"] translation_fallback: null fallback_fields: [] published_at: "2026-10-29T01:22:17.000Z" category: "Models" topics: ["openai","llm","deepseek"] source_urls: ["https://marketbrief.now/ai/chinese-ai-developers-use-openai-astra-to-train-new-models-following-ant-13684efc","https://huggingnews.com/ai/chinese-ai-developers-use-openai-astra-to-train-new-models-following-ant-13684efc"] summary: "Chinese AI developers DeepSeek and Moonshot are using outputs from OpenAI's Astra model to train new AI systems through a process called distillation, according to The Information. This shift follows Anthropic's implementation of restrictions on using its models for similar training practices, with Astra reportedly based on OpenAI's GPT-6 architecture." --- # Chinese AI Developers Use OpenAI Astra to Train New Models Following Anthropic Blocks > [Open the canonical story]() **Published:** 2026-10-29T01:22:17.000Z **Category:** Models **Topics:** openai, llm, deepseek ## Summary Chinese AI developers DeepSeek and Moonshot are using outputs from OpenAI's Astra model to train new AI systems through a process called distillation, according to The Information\. This shift follows Anthropic's implementation of restrictions on using its models for similar training practices, with Astra reportedly based on OpenAI's GPT\-6 architecture\. ## Sources - [Story source]() - [Story source]()