---
format: "aidr-story-markdown/v1"
id: "3c6d97d57f3f5a3cd9cca2199cd6dd4ecb634832532866b87fc882ec3beb2943"
canonical_url: "https://aidr.today/3c6d97d5?lang=en"
title: "Sources: DeepSeek and Moonshot AI heavily relied on GPT-6 Astra distillation to train models like Kimi K3.1 Pro and V4.1-Flash 1017, after Anthropic's crackdown (The Information)"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-10-29T10:46:03.000Z"
category: "Models"
topics: ["openai","llm"]
source_urls: ["https://www.techmeme.com/261029/p19"]
summary: "Sources say DeepSeek and Moonshot AI relied on GPT-6 Astra distillation to train models such as Kimi K3.1 Pro and V4.1-Flash 1017 following Anthropic's crackdown, with OpenAI's GPT-6 Astra seeing strong demand since its early September release."
---

# Sources: DeepSeek and Moonshot AI heavily relied on GPT\-6 Astra distillation to train models like Kimi K3\.1 Pro and V4\.1\-Flash 1017, after Anthropic's crackdown \(The Information\)

> [Open the canonical story](<https://aidr.today/3c6d97d5?lang=en>)

**Published:** 2026-10-29T10:46:03.000Z
**Category:** Models
**Topics:** openai, llm

## Summary

Sources say DeepSeek and Moonshot AI relied on GPT\-6 Astra distillation to train models such as Kimi K3\.1 Pro and V4\.1\-Flash 1017 following Anthropic's crackdown, with OpenAI's GPT\-6 Astra seeing strong demand since its early September release\.

## Sources

- [Story source](<https://www.techmeme.com/261029/p19>)