GLM-5.3-FlashX launches on B.AI at 200 tokens/sec
Z.AI's GLM-5.3-FlashX is live on B.AI's API and web chat, a 320B total/18B active sparse model with a 1M context window, native multimodality and up to 200 tokens/sec for interactive coding and agent workflows.
📢 GLM-5.3-FlashX Is Now Live on http://B.AI Developed by http://Z.AI, GLM-5.3-FlashX is a speed-optimized native multimodal model built on an efficient 320B total / 18B active parameter sparse architecture, reaching generation speeds of up to 200 tokens/sec. It integrates a 1M context window, native understanding of text/images/videos/files, and interleaved reasoning for high-speed interactive coding and agent workflows. Now available on both http://B.AI API and Web Chat! 👉 Try now: https://chat.b.ai/chat 🔗 Learn more: https://docs.b.ai/llmservice/models/glm-5-3-flashx/