77.8% SWE-Bench open-weight model with strong agentic and terminal capabilities.
Full-stack development, agentic tasks, structured reasoning
200K tokens
Access GLM-5 alongside 41 other AI models. Switch models per message — no lock-in.
Get Started Free