
DeepSeek V4 Model Preview: AI Race Heats Up
Chinese AI firm DeepSeek released a preview of its open-source Version 4 LLM on February 26, 2026, offering 'pro' and 'flash' versions. The company claims V4 excels in agent-based tasks and knowledge processing. This release follows previous milestones, including the V3 model and the highly successful R1 reasoning model. The announcement intensifies the U.S.-China tech rivalry, with reports suggesting early access was favored for domestic Chinese firms. DeepSeek continues to compete in a rapidly evolving and highly competitive Chinese AI market.






















