z-ai/glm-5.3-flash

Online Chat

智谱

Publish time

2026/8/27

Model Series

GLM

Input type

Output type

Input Price

¥0.76 / 1M tokens

Output Price

¥2.66 / 1M tokens

Cache Write Price

¥0.95 / 1M tokens

Cache Read Price

¥0.2185 / 1M tokens

Context Window

1,000,000

Max Output Length

128,000

GLM-5.3-Flash 是 GLM-5 系列首个原生多模态模型，以极致低成本架构实现超越 GLM-5.2 的更强智能： GLM-5.3-Flash 总参数量 320B、激活参数量 18B，是首个采用稀疏注意力与线性注意力混合架构的开源前沿模型。在保持精准长上下文能力的同时，显著降低计算与服务成本。相比 GLM-5.3，其注意力计算量和 KV 缓存大小分别降低 3.01 倍和 4.44 倍。视觉能力被原生融入 Coding 循环，使模型能够主动观察界面、渲染结果与交互反馈，并据此持续测试和改进。无论是前端开发、游戏构建、Blender 3D 场景，还是 BUA、CUA 驱动的真实环境操作，模型都能在代码、浏览器和图形界面之间协同完成任务。 GLM-5.3-Flash 进一步拓展至 Office、金融研究和专业文档等工作场景。它能够自主拆解复杂目标、调用合适工具、检查并优化输出，完成从研究分析、模型构建到 PPTX、PDF、DOCX、XLSX 成品交付的完整工作流。

Providers for z-ai/glm-5.3-flash

Zhinao API routes requests to the best-fit provider and automatically fails over to the one with highest availability.

Sort by

国内

TTFT

No data

Throughput

No data

Uptime

100.00%

Provider Model

nc/z-ai/glm-5.3-flash

Supported Parameters

temperaturetop_ptop_k

Recent Uptime

10月11日 5 PM100.00%

Reasoning

Toggleable

Supported Response Formats

OpenAI Chat CompletionsAnthropic Messages

Total Context

1,000,000

Max Output

128,000

Input Price-30%

¥0.56 / 1M tokens

Output Price-30%

¥1.96 / 1M tokens

Cache Write-30%

¥0.7 / 1M tokens

Cache Read-30%

¥0.161 / 1M tokens

paratera并行科技

国内

TTFT

4.86s

Throughput

5.87tps

Uptime

74.00%

Provider Model

paratera/z-ai/glm-5.3-flash

Supported Parameters

temperaturetop_ptop_k

Recent Uptime

10月11日 11 PM73.64%

Reasoning

Toggleable

Supported Response Formats

OpenAI Chat Completions

Total Context

1,000,000

Max Output

128,000

Input Price-10%

¥0.72 / 1M tokens

Output Price-10%

¥2.52 / 1M tokens

Cache Write-10%

¥0.9 / 1M tokens

Cache Read-10%

¥0.207 / 1M tokens

tokenmall-zyld

国内

TTFT

12.02s

Throughput

27.66tps

Uptime

100.00%

Provider Model

tokenmall-zyld/z-ai/glm-5.3-flash

Supported Parameters

temperaturetop_ptop_k

Recent Uptime

10月11日 11 PM100.00%

Reasoning

Toggleable

Supported Response Formats

OpenAI Chat CompletionsOpenAI ResponsesAnthropic Messages

Total Context

1,000,000

Max Output

128,000

Input Price-40%

¥0.48 / 1M tokens

Output Price-40%

¥1.68 / 1M tokens

Cache Write-40%

¥0.6 / 1M tokens

Cache Read-40%

¥0.138 / 1M tokens

智

智谱

国内

TTFT

7.64s

Throughput

29.72tps

Uptime

100.00%

Provider Model

bigmodel/z-ai/glm-5.3-flash

Supported Parameters

temperaturetop_ptop_k

Recent Uptime

10月11日 11 PM100.00%

Reasoning

Toggleable

Supported Response Formats

OpenAI Chat CompletionsOpenAI ResponsesAnthropic Messages

Total Context

1,000,000

Max Output

128,000

Input Price-5%

¥0.76 / 1M tokens

Output Price-5%

¥2.66 / 1M tokens

Cache Write-5%

¥0.95 / 1M tokens

Cache Read-5%

¥0.2185 / 1M tokens

火

火山引擎

国内

TTFT

3.82s

Throughput

50.63tps

Uptime

100.00%

Provider Model

volcengine/z-ai/glm-5.3-flash

Supported Parameters

temperaturetop_ptop_k

Recent Uptime

10月11日 11 PM100.00%

Reasoning

Toggleable

Supported Response Formats

OpenAI Chat CompletionsOpenAI Responses

Total Context

1,000,000

Max Output

128,000

Input Price

¥0.8 / 1M tokens

Output Price

¥2.8 / 1M tokens

Cache Write

¥1 / 1M tokens

Cache Read

¥0.23 / 1M tokens

Performance for z-ai/glm-5.3-flash

Compare different providers across Zhinao API

Throughput

29.77 tok/s

TTFT

7.04 s

Uptime for z-ai/glm-5.3-flash

Uptime for z-ai/glm-5.3-flash across all providers

Sample code and API for z-ai/glm-5.3-flash

Get API Key

Zhinao API normalizes requests and responses across providers for you

View full docs

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.360.cn/v1",
  apiKey: process.env.ZHINAO_API_KEY,
});

const response = await client.chat.completions.create({
  model: "z-ai/glm-5.3-flash",
  messages: [
    { role: "user", content: "Hello, how are you?" }
  ],
  temperature: 0.7,
  max_tokens: 1000,
});

console.log(response.choices[0].message.content);

z-ai/glm-5.3-flash

Online Chat

智谱

Publish time

2026/8/27

Model Series

GLM

Input type

Output type

Input Price

¥0.76 / 1M tokens

Output Price

¥2.66 / 1M tokens

Cache Write Price

¥0.95 / 1M tokens

Cache Read Price

¥0.2185 / 1M tokens

Context Window

1,000,000

Max Output Length

128,000

Providers for z-ai/glm-5.3-flash

Zhinao API routes requests to the best-fit provider and automatically fails over to the one with highest availability.

Sort by

国内

TTFT

No data

Throughput

No data

Uptime

100.00%

Provider Model

nc/z-ai/glm-5.3-flash

Supported Parameters

temperaturetop_ptop_k

Recent Uptime

10月11日 5 PM100.00%

Reasoning

Toggleable

Supported Response Formats

OpenAI Chat CompletionsAnthropic Messages

Total Context

1,000,000

Max Output

128,000

Input Price-30%

¥0.56 / 1M tokens

Output Price-30%

¥1.96 / 1M tokens

Cache Write-30%

¥0.7 / 1M tokens

Cache Read-30%

¥0.161 / 1M tokens

paratera并行科技

国内

TTFT

4.86s

Throughput

5.87tps

Uptime

74.00%

Provider Model

paratera/z-ai/glm-5.3-flash

Supported Parameters

temperaturetop_ptop_k

Recent Uptime

10月11日 11 PM73.64%

Reasoning

Toggleable

Supported Response Formats

OpenAI Chat Completions

Total Context

1,000,000

Max Output

128,000

Input Price-10%

¥0.72 / 1M tokens

Output Price-10%

¥2.52 / 1M tokens

Cache Write-10%

¥0.9 / 1M tokens

Cache Read-10%

¥0.207 / 1M tokens

tokenmall-zyld

国内

TTFT

12.02s

Throughput

27.66tps

Uptime

100.00%

Provider Model

tokenmall-zyld/z-ai/glm-5.3-flash

Supported Parameters

temperaturetop_ptop_k

Recent Uptime

10月11日 11 PM100.00%

Reasoning

Toggleable

Supported Response Formats

OpenAI Chat CompletionsOpenAI ResponsesAnthropic Messages

Total Context

1,000,000

Max Output

128,000

Input Price-40%

¥0.48 / 1M tokens

Output Price-40%

¥1.68 / 1M tokens

Cache Write-40%

¥0.6 / 1M tokens

Cache Read-40%

¥0.138 / 1M tokens

智

智谱

国内

TTFT

7.64s

Throughput

29.72tps

Uptime

100.00%

Provider Model

bigmodel/z-ai/glm-5.3-flash

Supported Parameters

temperaturetop_ptop_k

Recent Uptime

10月11日 11 PM100.00%

Reasoning

Toggleable

Supported Response Formats

OpenAI Chat CompletionsOpenAI ResponsesAnthropic Messages

Total Context

1,000,000

Max Output

128,000

Input Price-5%

¥0.76 / 1M tokens

Output Price-5%

¥2.66 / 1M tokens

Cache Write-5%

¥0.95 / 1M tokens

Cache Read-5%

¥0.2185 / 1M tokens

火

火山引擎

国内

TTFT

3.82s

Throughput

50.63tps

Uptime

100.00%

Provider Model

volcengine/z-ai/glm-5.3-flash

Supported Parameters

temperaturetop_ptop_k

Recent Uptime

10月11日 11 PM100.00%

Reasoning

Toggleable

Supported Response Formats

OpenAI Chat CompletionsOpenAI Responses

Total Context

1,000,000

Max Output

128,000

Input Price

¥0.8 / 1M tokens

Output Price

¥2.8 / 1M tokens

Cache Write

¥1 / 1M tokens

Cache Read

¥0.23 / 1M tokens

Performance for z-ai/glm-5.3-flash

Compare different providers across Zhinao API

Throughput

29.77 tok/s

TTFT

7.04 s

Uptime for z-ai/glm-5.3-flash

Uptime for z-ai/glm-5.3-flash across all providers

Sample code and API for z-ai/glm-5.3-flash

Get API Key

Zhinao API normalizes requests and responses across providers for you

View full docs

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.360.cn/v1",
  apiKey: process.env.ZHINAO_API_KEY,
});

const response = await client.chat.completions.create({
  model: "z-ai/glm-5.3-flash",
  messages: [
    { role: "user", content: "Hello, how are you?" }
  ],
  temperature: 0.7,
  max_tokens: 1000,
});

console.log(response.choices[0].message.content);