DeepSeek Glossary

Chinese AI lab, founded 2023 as a research arm of the quantitative hedge fund High-Flyer. Makers of the DeepSeek-V and DeepSeek-R model families, with substantial reasoning-model contributions.

DeepSeek is a Chinese AI lab founded in July 2023 by Liang Wenfeng as a research arm of the Hangzhou-based quantitative hedge fund High-Flyer. The lab is headquartered in Hangzhou and Beijing. It has released a series of open-weights large language models — most prominently DeepSeek-V3 (a 671B-parameter Mixture-of-Experts general-purpose model) and DeepSeek-R1 (one of the first widely-distributed open-weights *reasoning models *, released January 2025, that triggered substantial market reaction including a one-day drop in US tech-stock valuations).

For this Dictionary, DeepSeek matters at two levels. First, technically: DeepSeek’s models are credible options in the Sovereign Compute tier, especially for operators who need reasoning capability at open-weights cost. Second, strategically: DeepSeek is one of the most legible signals that the PRC’s AI ecosystem is producing frontier-adjacent capability with materially different cost structures from the US labs. The Dictionary’s Capability Overhang argument and the in-progress zhengming (正名) essay both treat DeepSeek as structurally important.

The lab’s relationship to the Chinese state — described in DeepSeek’s own statements as that of an independent private research lab — is the subject of ongoing analysis the Dictionary may eventually address in a future entry.

See also

• *Reasoning Model *
• Sovereign Compute
• Capability Overhang
• Open-Weights Inversion (covered inside Sovereign Compute)

Return to Dictionary All Entries (A–Z) For Students Other Writing