Pricing, Performance & Features Comparison
GLM-4.7-Flash is a 30B Mixture-of-Experts (MoE) reasoning model with approximately 3.6B active parameters, designed for local deployment with best-in-class performance for coding, agentic workflows, and chat. It supports a 200K context window and achieves open-source state-of-the-art scores on benchmarks like SWE-bench Verified and τ²-Bench, excelling particularly in frontend and backend development capabilities.
Kimi K2.5 is Moonshot's most intelligent and versatile model to date, featuring a native multimodal architecture that supports both visual and text input alongside thinking and non-thinking modes. It achieves state-of-the-art performance in coding, reasoning, and Agent tasks, utilizing a 256K context window to solve complex logical and mathematical problems.