DeepSeek Launches New AI Model V4 on Huawei Chips, Competing with US Rivals

• •DeepSeek V4 model launched, optimized for Huawei's Ascend chips.
• •V4 supports ultra-long context lengths and claims to rival US AI models.
• •The model is open-source, promoting widespread adoption in various sectors.

On April 24, 2026, Chinese startup DeepSeek released its new AI model, V4, optimized for Huawei's Ascend chips, marking a significant advancement in China's AI capabilities. The model is available in two versions: V4-Pro with 1.6 trillion parameters and V4-Flash with 284 billion parameters. DeepSeek claims V4 can rival leading US models while dramatically reducing inference costs. The model supports an ultra-long context length of one million tokens, enhancing its reasoning capabilities. Analysts suggest that while V4 continues the trend of competitive Chinese AI, it may not generate the same market shock as its predecessor, the R1 model. DeepSeek's collaboration with Huawei reflects a strategic shift towards domestic chip reliance due to US export controls. The model is open-source, allowing widespread adoption across various sectors. The release is seen as a critical step in China's goal of self-sufficiency in AI technology.