GLM-4.5-Air is the lightweight variant of our latest flagship model family, also purpose-built for agent-centric applications. Like GLM-4.5, it adopts the Mixture-of-Experts (MoE) architecture but with a more compact parameter size. GLM-4.5-Air also supports hybrid inference modes, offering a "thinking mode" for advanced reasoning and tool use, and a "non-thinking mode" for real-time interaction. Users can control the reasoning behaviour with the `reasoning` `enabled` boolean. [Learn more in our docs](https://openrouter.ai/docs/use-cases/reasoning-tokens#enable-reasoning-with-default-config)

GLM-4.5 is our latest flagship foundation model, purpose-built for agent-based applications. It leverages a Mixture-of-Experts (MoE) architecture and supports a context length of up to 128k tokens. GLM-4.5 delivers significantly enhanced capabilities in reasoning, code generation, and agent alignment. It supports a hybrid inference mode with two options, a "thinking mode" designed for complex reasoning and tool use, and a "non-thinking mode" optimized for instant responses. Users can control the reasoning behaviour with the `reasoning` `enabled` boolean.

GLM-4.6 is the flagship model from Zhishen, their latest. It has a total of 355 billion parameters and an active parameter of 32 billion. GLM-4.6 has surpassed all core capabilities of GLM-4.5, specifically: Advanced coding ability: In both public benchmarks and real programming tasks, GLM-4.6's coding ability matches Claude Sonnet 4, making it the best Coding model domestically known. Context length: The context window has increased from 128K to 200K, allowing it to handle longer code and intelligent agent tasks. Inference ability: There has been an improvement in inference capabilities, and it now supports calling tools during the inference process. Search capability: Enhanced the model's performance in tool calling and search intelligent agents, performing better within the intelligent agent framework. Writing ability: In terms of style, readability, and role-playing scenarios, it more closely aligns with human preferences. Multilingual translation: Further strengthened the model's ability to handle cross-linguistic tasks. If you need any further assistance or have more text to translate, feel free to let me know!