China’s MiniMax releases cheap AI model ‘designed for real-world productivity’
Company says it is ‘first frontier model … delivering on the promise of intelligence too cheap to meter’
Chinese artificial intelligence (AI) company MiniMax has released a new model “designed for real-world productivity”, hot on the heels of new releases from its domestic rivals in a frenetic week for China’s AI industry.
Citing in-house benchmark scores, MiniMax said on Friday morning that M2.5 was its most powerful model yet, with performance matching leading models from the likes of American heavyweights Anthropic and OpenAI in key areas such as coding and search.
The new model was met positively by the markets on Friday, with the company’s shares in Hong Kong closing 15.7 per cent higher at HK$680.
One of the model’s strengths was its fast and cheap speed due to the cost efficiency achieved at a model size of 230 billion parameters, unchanged from previous M2 model iterations, MiniMax said.
“M2.5 is the first frontier model where users do not need to worry about cost, delivering on the promise of intelligence too cheap to meter,” the company said, adding that the model could be used continuously for an hour at a rate of 100 tokens a second for just US$1. Tokens are small chunks of data read or generated by a model.