Tag
The user tested Qwen3.7-Max and believes it matches top models like Claude 4.6 and Gemini 3.1 Pro in frontend, computing power, and Agent capabilities. Its reasoning ability has significantly improved, and with monthly iteration speed, it has become a first-tier domestic model.
An analysis of DeepSeek's long-term strategy, arguing that their innovations in MoE, GRPO, and KV cache reduction are aimed at building a 10T USD Chinese AI hardware ecosystem rather than selling immediate applications, potentially achieving a 1T USD valuation.
BitCPM-CANN is the first open-source 1.58-bit ternary LLM trained entirely on Chinese-developed AI infrastructure (Huawei Ascend 910B), offering extreme memory reduction for edge deployment.
Qwen 3.7 is an impressive new AI model from Chinese labs, with discussion on whether weights will be available for download.
A local-first AI slide deck generator that works offline with your own model, supporting Chinese AI services and HTML-based presentations.
Matthew White shares firsthand insights from visits to major Chinese AI labs and startups across Beijing, Shanghai, and Hangzhou, including observations on open model releases, robotics, and the policy landscape.
This article tests four open-source Chinese AI models — Zhipu GLM 5.1, Moonshot Kimi K2.6, Stepfun MIMO 2.5 Pro, and DeepSeek V4 Pro — on programming tasks. It finds that GLM leads overall in most tasks but not absolutely; each model has its own strengths and weaknesses.
DeepSeek secured a $7 billion funding round, the largest in Chinese AI history, backed by government funds and its founder, alongside massive integration into Chinese government, healthcare, and public services under national policy.
Zhang Xiaojun's podcast has become an important platform for China's top AI scientists to showcase research results and exchange ideas, reflecting the rising influence of media in the Chinese AI community.
The author provides a detailed look at Kimi's latest internal beta features — Claw Groups and Agent Clusters. Claw Groups allow multiple AIs to take on distinct roles in a group chat while challenging each other's outputs, while Agent Clusters can break down complex tasks and distribute them across 10 parallel sub-agents. The author used these features for investment research on tech stocks like NVIDIA, and sees this as a sign that AI tools have officially entered the "organizational" tier.