Microsoft for Startups Founders
AWS Activate Startup
IBM Business Partner
Edge Impulse Experts Network
Intel Software Innovators
Google cloud Startup
Supported by Business Wales
Supported by Enterprise Hub

DeepSeek

DeepSeek is a cutting-edge Chinese artificial intelligence company founded in July 2023 by Liang Wenfeng, dedicated to developing advanced large language models (LLMs). Based in Hangzhou, Zhejiang, the company has rapidly gained attention for its innovative approach to AI technology and cost-effective model training.

The company distinguishes itself through strategic hiring practices that prioritize technical abilities and diverse knowledge perspectives. By recruiting researchers from top Chinese universities and professionals outside traditional computer science backgrounds, DeepSeek cultivates a unique talent pool that enhances its AI model's versatility.

DeepSeek's technological prowess is underscored by its impressive computational infrastructure, including sophisticated computing clusters like Fire-Flyer and Fire-Flyer 2. These advanced systems enable the company to develop high-performance AI models with remarkable efficiency.

  • Founded in July 2023 by Liang Wenfeng.
  • Develops large language models with significantly lower training costs.
  • Operates advanced computing clusters with over 5,000 GPUs.
  • Focuses on research and innovative AI development.
  • Recruits diverse talent to expand AI knowledge boundaries.

Platforms

Models

DeepSeek R1 Converse

Powered by large-scale reinforcement learning, DeepSeek R1 Converse delivers intelligent dialogue experiences through state-of-the-art reasoning models that dynamically adapt to complex communication challenges across technical and linguistic domains.

  • Context Window: 128,000
  • TPM: 200,000
  • RPM: 200
  • Embedding Size: NA
Attributes

Conversation, Reasoning

DeepSeek R1 Text

DeepSeek R1 Text delivers advanced text generation capabilities, utilizing reinforcement learning to produce precise, coherent content across English and Chinese languages. Engineered for complex reasoning tasks, this model excels in generating high-quality text with remarkable accuracy and contextual understanding.

  • Context Window: 128,000
  • TPM: 200,000
  • RPM: 200
  • Embedding Size: NA
Attributes

Text generation, Code generation

6LfEEZcpAAAAAC84WZ_GBX2qO6dYAEXameYWeTpF