Often discuss my three-level vision for opening GLM to the community: First, we focus on accessibility by lowering the barrier to entry and …

Often discuss my three-level vision for opening GLM to the community: First, we focus on accessibility by lowering the barrier to entry and removing unnecessary constraints so developers can truly explore the model. Second, we provide a robust baseline that empowers everyone to fine-tune, quantize, and adapt the weights to their own unique use cases. Finally, we aim to contribute to the collective norm by sharing our insights on model architecture, training methodologies, and data curation. We'd like to help define how the next generation of open models could be built. Silicon Valley is quietly running on Chinese open source AI models. Here are the receipts: → Cursor confirmed last month that Composer 2 is built on Moonshot's Kimi K2.5 → Cognition's SWE-1.6 model is likely post-trained on Zhipu's GLM → Shopify saved $5M a year by switching to A…

We're seeing even more autonomous AI coworkers. The new MLE agent on the market is Disarray. In Kaggle competitions, Disarray: - won 28 meda…
New course: Efficient Inference with SGLang: Text and Image Generation, built in partnership with LMSys @lmsysorg and RadixArk @radixark, an…
🫡 @aiDotEngineer in London was extremely good. Absolutely incredible speaker line up, great curation, escaped the Silicon Valley bubble mas…
The Information wrote that Meta consumes 60.2T tokens per month, or ~750M per employee. Meta is getting absolutely token-mogged by us curren…
This is literally why I wrote Taming Silicon Valley.

Source: Zhipu AI (X) | 2026-04-10

Related