GLM-5.2 draws Silicon Valley attention

GLM-5.2, a new open-source model from China’s z.AI, is drawing attention across Silicon Valley for long coding tasks and agentic workflows. The company says it operates on a 1 million token context window, putting it in the same league as Anthropic’s Claude Opus 4.8 and OpenAI’s GPT 5.5.

Tech executives and investors praised the model’s performance on social media after its launch last week. Vercel CEO Guillermo Rauch said he was “genuinely impressed” by its coding ability, while former Meta, Google DeepMind, and Microsoft executive Matt Velloso called it the first open model that met his bar as a daily driver.

The response echoes the reaction to DeepSeek’s R1, which raised doubts about the durability of Silicon Valley’s AI lead. Open models can be downloaded, modified, and run inside a company’s own systems, unlike closed models from companies such as OpenAI and Anthropic. Anthropic recently warned that China is narrowing the gap through looser chip controls and distillation attacks, while saying the US and allies may still be able to “lock in a 12-24 month lead in frontier capabilities.”