The Neural Feed Pulse

DeepSeek Upgrades DeepSeek-Reasoner Model to DeepSeek-R1-0528 with Enhanced Capabilities

🗃 Viral Wire ⚡ Analysis

DeepSeek upgraded its reasoner to DeepSeek-R1-0528, boosting AIME 2025 Pass@1 by 17.5 points (70.0 to 87.5). This reduces hallucinations but increases token con

📖 Read Full Pulse