📰 DeepSeek in the News – April 2025 Highlights

A roundup of key developments surrounding DeepSeek and the growing AI race in China, featuring breakthroughs, model launches, and rival competition.

---

🇰🇷 DeepSeek Returns to South Korea After Suspension

🗓️ Date: 2025-04-28
🔗 Read the full article

DeepSeek is once again available for download in South Korea after a two-month suspension due to data protection violations. This marks a critical step in its return to international markets.

---

🚀 DeepSeek R2 Rumored to Launch Soon

🗓️ Date: 2025-04-28
🔗 Read the full article

Speculation is building around the release of DeepSeek R2, a next-generation reasoning model expected to challenge OpenAI’s latest models like o3 and o4-mini.

---

🇨🇳 Alibaba Targets DeepSeek with Upgraded LLM

🗓️ Date: 2025-04-29
🔗 Read the full article

Alibaba released a major upgrade to its large language model, intensifying the competition with DeepSeek and Baidu as tech giants battle for dominance in China’s AI space.

---

🧠 AI Model Wars Heat Up Ahead of Golden Week

🗓️ Date: 2025-04-30
🔗 Read the full article

As China prepares for its Labor Day Golden Week, DeepSeek, Baidu, and Alibaba are all releasing new LLMs, accelerating the arms race in generative AI across the country.

---

📱 Xiaomi Enters AI Arena with DeepSeek-Inspired Model

🗓️ Date: 2025-04-29
🔗 Read the full article

Xiaomi has launched its own open-source AI model, MiMo, a reasoning system that mirrors the capabilities of DeepSeek’s R1. This signals Xiaomi’s official entry into the Chinese AI ecosystem.

---

🔬 DeepSeek R2 Rumored Specs and Training Strategy

🗓️ Date: 2025-04-27
🔗 Read the full article

Leaks suggest DeepSeek R2 will feature 1.2 trillion parameters and a hybrid MoE architecture. The model is said to be 97% cheaper to train than GPT-4, using Huawei’s Ascend 910B chips for high-efficiency enterprise AI.

---