From DeepSeek to Deep Research and beyond

DeepSeek, the Chinese AI company that released its open source DeepSeek V3 LLM (a competitor to OpenAI GPT-4o and others) at the end of 2024 recently released its similarly open sourced (well, mostly open sourced given that model weights are shared, but not the underlying training datasets/code) reasoning-based LLMs DeepSeek-R1-Zero and DeepSeek-R1. These compete with OpenAI’s o3 proprietary LLM and others (e.g., Anthropic’s Claude 3.5 Sonnet).

DeepSeek is just the latest episode of the AI arms race towards AGI and AI agents. There has been a market and investor fascination with V3 (and R1 variants), but what does R1 know about procurement? Let's explore ...