GLM-5.3-FlashX is Z.ai's speed-optimized variant of GLM-5.3-Flash, a natively multimodal MoE model (around 320B total / 18B active parameters, MIT-licensed, 1M-token context) built on the same hybrid sparse-and-linear attention architecture. It generates at up to 200 tokens/sec for latency-sensitive coding and long-horizon agent tasks that still need the full 1M-token context, at a premium over the standard GLM-5.3-Flash endpoint.
费用比率
0.3x
上下文
1M
发布日期
2026年9月18日
输入
TextImageVideo
输出
Text
支持
推理工具调用结构化输出
关于 GLM 5.3 FlashX 的最新资讯
近期文章和社区帖子,均链接至原始来源。
- Fixstars adds GLM-5.3 to its Japan-hosted AI inference serviceInternational Business Times


GLM-5.3 Flash is FREE! 3 Ways to Use It (OpenCode, NVIDIA & Z.ai)Code With Yousaf14,936
Kimi K3.1 Pro vs GLM 5.3: 10-Round SHOWDOWNLifey Guy1,362
ZCode's Free GLM-5.3-Flash Tokens: How to ClaimOkay, but simple?1,591- GLM-5.3-FlashHacker News1,132580
- GLM-5.3: Frontier coding with emergent cyber capabilitiesHacker News1,171584