DeepSeek and Kimi reportedly trained on answers from OpenAI's top AI The two Chinese labs have relied heavily on outputs from GPT-6 Astra to build their newest models, The Information reported on Wednesday, citing people with direct knowledge. The trick is called refinement: a lab feeds a more powerful model its own answers, so it learns the skills much more cheaply. The named models are Kimi K3.1 Pro Moonshot, released last week, and DeepSeeks V4.1-Flash 1017, released since October 17th. OpenAI, DeepSeek and Moonshot did not respond to Reuters requests for comment. Claude had trouble copying this month. Anthropic said new restrictions, account bans and blocks cut the data distillers harvest by about 90% to 99%. OpenAI said on September 30 that it tied a refining campaign that began in July to people associated with the moonshot.

DeepSeek and Kimi reportedly trained on answers from OpenAI's top AI The two Chinese labs have relied heavily on outputs from GPT-6 Astra to build their newest models, The Information reported on Wednesday, citing people with direct knowledge. The trick is called refinement: a lab feeds a more powerful model its own answers, so it learns the skills much more cheaply. The named models are Kimi K3.1 Pro Moonshot, released last week, and DeepSeeks V4.1-Flash 1017, released since October 17th. OpenAI, DeepSeek and Moonshot did not respond to Reuters requests for comment. Claude had trouble copying this month. Anthropic said new restrictions, account bans and blocks cut the data distillers harvest by about 90% to 99%. OpenAI said on September 30 that it tied a refining campaign that began in July to people associated with the moonshot.