DeepSeek's Prover V2: The 671-Billion Parameter AI Model Challenging OpenAI
DeepSeek, a rising star in the AI industry, has set tongues wagging with their latest launch, Prover V2—a 671-billion parameter AI model designed to verify mathematical theorems. Positioned as a contender against US-based OpenAI, Prover V2 uses Lean 4 for robust proof verification and is made hardware-friendly through 8-bit floating point quantization. Released under an open-access MIT license, this model is stirring debates over scientific progress versus its potential for misuse, echoing the global movement toward more accessible AI tools.
May 2