Chinese AI labs reportedly shift distillation target to OpenAI's Astra
What happened
On October 29, The Decoder reported that Chinese AI labs have shifted their main distillation target from Claude to OpenAI's GPT-6 Astra after Anthropic tightened its safeguards, using Astra's outputs to train models. Citing people familiar with the matter via The Information, the report says DeepSeek and Moonshot AI make heavy use of Astra outputs, involving Kimi K3.1 Pro and V4.1-Flash 1017. Such distillation depends on obtaining hidden reasoning steps, which both OpenAI and Anthropic prohibit. The usage remains an account from unnamed sources in the report.
Written by AI from the coverage · updated 2 hours ago
Coverage
Follow the reports to see the story from different sides.
- The DecoderOpenAI's Astra becomes the main target for Chinese AI labs after Anthropic tightens its defenses
中国 AI 实验室据报将主要蒸馏目标从 Claude 转向 OpenAI 的 GPT-6 Astra，利用其输出训练模型。The Information 援引知情人士称，DeepSeek 和月之暗面大量使用 Astra 输出，涉及 Kimi K3.1 Pro 和 V4.1-Flash 1017；这类蒸馏依赖获取隐藏推理步骤，OpenAI 和 Anthropic 均禁止此行为。
Heat over time
Not enough continuous observations to draw a trend yet.