GBAF Logo
Global Banking & Finance Awards® 2026 Nominations open, free to enter Nominate now →
DeepSeek, Moonshot used OpenAI's GPT-6 Astra to train latest models, The Information reports - Finance news and analysis from Global Banking & Finance Review
Finance

DeepSeek, Moonshot used OpenAI's GPT-6 Astra to train latest models, The Information reports

Published by Global Banking & Finance Review

Posted on October 29, 2026

2 min read

· Last updated: October 29, 2026

Add as preferred source on Google

DeepSeek and Moonshot AI Use OpenAI GPT-6 Astra in Latest Model Training

Chinese AI Developers Leverage GPT-6 Astra for Model Advancement

Background on Model Training Practices

Oct 28 (Reuters) - Chinese AI developers DeepSeek and Moonshot AI relied heavily on outputs from OpenAI's powerful GPT-6 Astra model to train their latest models, The Information reported on Wednesday, citing people with direct knowledge of the matter.

The Rise of Distillation in AI Development

• The practice, known as distillation, has become a flashpoint for leading AI companies as US-China technology competition intensifies, with firms arguing it allows rivals to reproduce advanced capabilities more quickly and cheaply

Recent Model Releases and Their Benefits

• Moonshot AI's Kimi K3.1 Pro model, which was released last week, and DeepSeek's upgraded V4.1-Flash 1017 model, which was launched on October 17, benefited from distilling Astra, The Information said

Industry Reactions and Security Measures

• OpenAI, DeepSeek and Moonshot AI did not immediately respond to Reuters requests for comment

OpenAI's Response to Distillation Campaigns

• In a September 30 blog post, OpenAI said it had linked a core part of a distillation campaign that began in July to individuals associated with Moonshot AI

Perspectives from Leading AI Firms

• Firms including Anthropic, OpenAI and Google have argued the practice enables rivals to reproduce advanced capabilities more quickly and at lower cost

Anthropic's Preventative Measures

• Earlier this month, Anthropic said new measures designed to prevent rivals from copying the capabilities of its Claude AI models have reduced data harvested in so-called distillation attacks by about 90% to 99%

(Reporting by Disha Mishra in Bengaluru; Editing by Sherry Jacob-Phillips and Subhranshu Sahu)

Key Takeaways

  • DeepSeek and Moonshot leveraged GPT‑6 Astra outputs through distillation to improve performance rapidly and cost‑effectively (theinformation.com)
  • Distillation is controversial: OpenAI and Anthropic have warned it lets rivals reproduce capabilities cheaply, with Anthropic claiming its new safeguards cut data harvesting by 90–99% (theinformation.com)
  • U.S. security agencies accuse several Chinese AI firms, including DeepSeek and Moonshot, of conducting industrial‑scale distillation campaigns to bridge technology gaps and evade export controls (investing.com)

References

Frequently Asked Questions

What AI model did DeepSeek and Moonshot AI use for training?
DeepSeek and Moonshot AI relied on outputs from OpenAI's GPT-6 Astra model to train their latest AI models.
What is AI model distillation?
AI model distillation is a process where a model is trained using the outputs from another, more advanced model to reproduce its capabilities more quickly and cheaply.
Why is the practice of distillation controversial in AI development?
Distillation has become controversial because it enables rivals to rapidly reproduce advanced capabilities, intensifying technology competition between countries like the US and China.
Which models benefited from distilling GPT-6 Astra?
Moonshot AI's Kimi K3.1 Pro and DeepSeek's upgraded V4.1-Flash 1017 models benefited from distilling GPT-6 Astra.
What measures have AI firms taken to prevent distillation?
Firms like Anthropic have introduced measures to prevent rivals from copying capabilities, reducing data harvested in distillation attacks by up to 99%.

Tags

Related Articles

More from Finance

Explore more articles in the Finance category