GLM 5.3 FlashX now available on AI Gateway

Vercel added GLM 5.3 FlashX, Z.ai's multimodal coding model, to its AI Gateway, offering inference at roughly 200 tokens per second. The high-speed serving option is aimed at coding agents, tool loops, and interactive applications, and is callable as zai/glm-5.3-flashx across API formats. Vercel said AI Gateway reflects provider pricing with no markup and charges no platform fee on inference, including Bring Your Own Key requests.

GLM 5.3 FlashX https://vercel.com/ai-gateway/models/glm-5.3-flashx is now available on AI Gateway. GLM 5.3 FlashX is a high-speed serving option for Z.ai's multimodal coding model, delivering inference at ~200 tokens per second for faster streamed responses. The higher serving speed is useful for coding agents, tool loops, and interactive applications where users wait on generated output. Use zai/glm-5.3-flashx across API formats and in coding agents: To use it in a coding agent, see the coding agents guide https://vercel.com/docs/ai-gateway/coding-agents , then run vercel ai-gateway setup to create a key and configure your supported agents. Select zai/glm-5.3-flashx inside the agent. AI Gateway provides a unified API for calling models, tracking usage and cost, and configuring retries, failover, and performance optimizations for higher-than-provider uptime. It includes built-in custom reporting https://vercel.com/changelog/custom-reporting-ai-gateway , budgets for API keys https://vercel.com/docs/ai-gateway/authentication-and-byok/api-keys , routing rules https://vercel.com/docs/ai-gateway/models-and-providers/routing-rules , and more. AI Gateway reflects provider pricing with no markup and does not charge a platform fee on inference, including on Bring Your Own Key https://vercel.com/docs/ai-gateway/authentication-and-byok/byok BYOK requests. Try GLM-5.3-FlashX in the model playground https://vercel.com/ai-gateway/models/glm-5.3-flashx , or view all language models https://vercel.com/ai-gateway/models?type=text available on AI Gateway.