GLM-5.3-Flash

GLM-5 系列首个原生多模态模型，总参数 320B，激活参数 18B，低成本。

Zhipu AIglm-5.3-flash2026-08-26报告数据错误

规格

上下文: 1.05M
最大输出: 131.1K
输入价: $0.15/ 百万 Token
输出价: $0.5/ 百万 Token
发布日期: 2026-08-26

上下文

上下文: 1.05M
最大输入: —
最大输出: 131.1K

价格 · USD / 百万 Token

输入: $0.15
输出: $0.5
缓存读取: $0.03
缓存写入: —

输入输出

输入: 文本图像视频文档
输出: 文本

API

接口类型: chat

思考

支持思考: 支持
思考等级: —

源信息

状态: 可用
发布日期: 2026-08-26
知识截止: —

能力

已确认: toolsstructured_outputstreamingcaching

调用方式

zhipumodel = glm-5.3-flashreasoning = thinking.type

chat

POSThttps://api.z.ai/api/paas/v4/chat/completions

JSON

glm-5.3-flash.json

{
  "id": "glm-5.3-flash",
  "object": "model",
  "created": 1787702400,
  "owned_by": "zhipu",
  "name": "GLM-5.3-Flash",
  "api": {
    "types": [
      "chat"
    ]
  },
  "limits": {
    "context": 1048576,
    "input": null,
    "output": 131072
  },
  "modalities": {
    "input": [
      "text",
      "image",
      "video",
      "document"
    ],
    "output": [
      "text"
    ]
  },
  "reasoning": {
    "supported": true,
    "efforts": []
  },
  "pricing": {
    "currency": "USD",
    "unit": "1M_tokens",
    "input": 0.15,
    "output": 0.5,
    "cache_read": 0.03,
    "cache_write": null
  },
  "features": [
    "tools",
    "structured_output",
    "streaming",
    "caching"
  ],
  "info": {
    "status": "active",
    "release_date": "2026-08-26",
    "knowledge_cutoff": null,
    "description": "First native multimodal model in the GLM-5 series, 320B total and 18B active parameters, low cost.",
    "docs": "https://docs.z.ai/guides/vlm/glm-5.3-flash",
    "verified_at": "2026-10-10"
  }
}

官方来源

GLM-5.3-Flash | Z.AI Developer Documentdocs.z.ai2026-10-10 Pricing | Z.AI Developer Documentdocs.z.ai2026-10-10 Release Notes | Z.AI Developer Documentdocs.z.ai2026-10-10 Create Chat Completion (max_tokens maximum 131072) | Z.AI Developer Documentdocs.z.ai2026-10-10 GLM-4.6 config.json (max_position_embeddings 202752) | Hugging Facehuggingface.co2026-10-10