Z.ai's GLM-5.2, a 753-billion-parameter model, outperforms GPT-5.5 on software engineering benchmarks with a 1-million-token context window. The model is available on Hugging Face, Z.ai API, and 20+ coding platforms. It operates at one-sixth the cost of GPT-5.5. The release advances open-source capabilities in long-horizon coding tasks.
Tap to vote and see what everyone thinks.
Summary by ByteBrief
GLM-5.2 tops frontend coding benchmarks