ai
Chinese AI Model GLM-5.2 Draws Silicon Valley Attention for Coding Tasks
z.AI released GLM-5.2 last week, an open-source model with a 1 million token context window. Industry figures posted reactions on X highlighting its performance on long coding tasks. The release follows earlier open-source models from Chinese developers.
Audio version
Tap play to generate a narrated version.
Developing·Limited corroboration so far. This page will refresh as more sources emerge.
The company states the model operates on a 1 million token context window. Vercel CEO Guillermo Rauch posted on X that he was "Genuinely impressed, almost shocked, at how good GLM-5.2 by @zai_org is at coding. " Matt Velloso, a former vice president at Meta, Google DeepMind, and Microsoft, wrote on X that he spent a full day using the model.
Most leading American models from OpenAI and Anthropic remain closed, requiring users to access them through company servers. The company positions the model for long-running coding tasks, placing its context window size alongside recent releases from Anthropic and OpenAI.
The release follows DeepSeek's R1 model from January 2025, which also drew attention for its capabilities at lower cost. U.S. export controls on advanced chips remain in place while Chinese developers continue releasing open-source alternatives. Anthropic stated in a recent report that China is narrowing the gap through looser chip access and distillation techniques, while noting the U.S. and allies still hold a lead in frontier capabilities.
