Insights
Forget the price wars: MiniMax goes open-source to rewrite the AI playbook
Like DeepSeek, MiniMax has also open-sourced the latest of its AI tech.
10 mins read
Foundation and Large Models
Even Microsoft, a key OpenAI backer, is embracing DeepSeek’s latest model as industries look for alternatives.
The Chinese startup’s latest release proves that expensive hardware isn’t the only route to AI innovation, shaking up markets worldwide.
36Kr investigates how natural language processing is evolving in China, driven by tech innovation, policy support, and industry needs.
Slated for a 2025 release, this new training method promises to elevate AI model precision and efficiency for specialized fields.
The AI veteran discusses the industry’s evolution, highlighting opportunities in cost-efficient models and the race to turn innovation into impact.
SenseTime spins off non-core segments into standalone units in pursuit of profitability.
Yang Zhilin, founder of Moonshot AI, also unveiled K0-math, a new model designed for mathematical reasoning.
As open-source models and high costs squeeze the market, leading Chinese AI firms consider dropping pre-trained models for a clearer path to profitability.
