tadamichi
Chinese AI startup MiniMax is making waves after it claimed that its new reasoning large language model called MiniMax-M1 was even better than DeepSeek’s (DEEPSEEK) upgraded AI model R1.
MiniMax, a Shanghai-based company, is part of the so called ‘new AI tigers’ of China which include Moonshot AI, Baichuan and Zhipu AI.
The company was founded by Yan Junjie in 2021 who is also the CEO.
MiniMax is backed by Chinese e-commerce and tech giant Alibaba and gaming company Tencent (OTCPK:TCEHY) (OTCPK:TCTZF). Alibaba continues to bet heavily on the AI sector and has been one of China’s most high-profile AI venture investors, with stakes in all four ‘new AI tigers.’
Last year in March it was reported that Alibaba was leading a funding round of at least $600M for MiniMax. The Chinese startup secured funds from Alibaba and other investors at a valuation of over $2.5B.
On Tuesday, MiniMax claimed that its new model outdoes all closed-source competitors from China in complex scenarios for productivity. M1 also scored higher than DeepSeek’s R1-0528 model.
The Chinese company used large-scale reinforcement learning to train M1, using 512 Nvidia (NVDA) H800 GPUs with a rental cost of $534,700, as per the statement. Earlier this year, DeepSeek rattled the U.S. tech sector after it unveiled its AI model R1 and detailed in a paper how LLMs could be built on tight budgets and improved without human supervision.
Even Apple’s (AAPL) CEO Tim Cook, who was on a visit to China in March, had praised DeepSeek. Last month, DeepSeek had upgraded its AI model R1.
More on tech stocks
- Alibaba: A Tech And AI Giant Severely Mispriced As A Retailer
- Alibaba: Strong Cloud And AI Progress Despite Poor Sentiment
- Alibaba Remains A Value Buy Here – Rich Upsides Ahead
- Alibaba’s update of Qwen3 models likely heralds Apple Intelligence coming to China: report
- China’s MiniMax claims new AI reasoning model surpasses DeepSeek