GLM-5.3-Flash

glm-5.3-flash

Open sourceGLM seriesNEW

Overview

Multimodal1M ContextVisionCode generationAgentic tasks

GLM-5.3-Flash is the first natively multimodal model in the GLM-5 series. Built on a 320B-A18B hybrid of sparse and linear attention, it is the world's first open frontier model to adopt that architecture. It scores 57 on the Artificial Analysis Intelligence Index, matching Claude Opus 4.8 and surpassing GLM-5.2 across the board. Vision is woven natively into the coding loop, so the model watches interfaces, rendered output and interaction feedback and keeps improving, working across code, browser and GUI to finish a task. It also reaches into Office work, financial research and professional documents, decomposing goals on its own, calling tools and delivering finished PPTX, PDF, DOCX and XLSX files. At one tenth the price of GLM-5.3, frontier intelligence no longer has to be rationed.

Radically cheaper architecture

A sparse-plus-linear attention hybrid cuts attention compute by 3.01× and KV cache by 4.44× versus GLM-5.3.

Natively multimodal visual coding

Vision is built into the coding loop: the model watches the interface, the rendered result and the interaction feedback, then keeps improving.

A work partner beyond coding

Extends to Office work, financial research and professional documents, delivering finished PPTX / PDF / DOCX / XLSX files.

Million-token context

A 1M-token context window with 128K max output loads project-scale codebases and very long documents in a single pass.

Features

VisionImages / video / files✓

Reasoninglow / high / max, max recommended; always on and cannot be disabled✓

Streaming✓

Tool callingIntegrates with a range of external tools✓

Context cachingImproves performance on long conversations✓

Structured outputJSON and other formats✓

Web search✕

Batch✕

Prefix completion✕

Fine-tuning✕

Pricing

Input¥0.8/M tokens

Input (cache hit)¥0.23/M tokens

Output¥2.8/M tokens

List prices are one tenth of GLM-5.3, matching the vendor's official pricing. A limited-time 50% discount runs until 2026-09-09 24:00 (input ¥0.4 / cached input ¥0.115 / output ¥1.4). Final billing follows the console and actual token usage.

Rate Limits & Context

Context Window1M

Max Output128K

RPM (requests/min)200

TPM (tokens/min)3M

API Reference

ENDPOINT

POST https://api.tokenfab.cn/v1/chat/completions

Request Example

curl https://api.tokenfab.cn/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $TOKENFAB_API_KEY" \
  -d '{
    "model": "glm-5.3-flash",
    "thinking": { "type": "enabled", "clear_thinking": false },
    "reasoning_effort": "max",
    "temperature": 1,
    "top_p": 0.95,
    "messages": [
      {"role": "system", "content": "You are a helpful assistant."},
      {"role": "user", "content": [
        {"type": "text", "text": "请分析这张界面截图，指出布局与交互问题并给出改进建议。"},
        {"type": "image_url", "image_url": {"url": "https://cdn.example.com/ui.png"}}
      ]}
    ]
  }'

from openai import OpenAI

client = OpenAI(
    api_key="your-api-key",
    base_url="https://api.tokenfab.cn/v1"
)

response = client.chat.completions.create(
    model="glm-5.3-flash",
    thinking={"type": "enabled", "clear_thinking": False},
    reasoning_effort="max",
    temperature=1,
    top_p=0.95,
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": [
            {"type": "text", "text": "请分析这张界面截图，指出布局与交互问题并给出改进建议。"},
            {"type": "image_url", "image_url": {"url": "https://cdn.example.com/ui.png"}}
        ]}
    ]
)

print(response.choices[0].message.content)

import OpenAI from "openai";

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

const response = await client.chat.completions.create({
  model: "glm-5.3-flash",
  thinking: { type: "enabled", clear_thinking: false },
  reasoning_effort: "max",
  temperature: 1,
  top_p: 0.95,
  messages: [
    { role: "system", content: "You are a helpful assistant." },
    { role: "user", content: [
      { type: "text", text: "请分析这张界面截图，指出布局与交互问题并给出改进建议。" },
      { type: "image_url", image_url: { url: "https://cdn.example.com/ui.png" } }
    ]}
  ]
});

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

Response Example

JSON

{
  "id": "chatcmpl-glm53f-8e3j9a2",
  "object": "chat.completion",
  "created": 1784950000,
  "model": "glm-5.3-flash",
  "choices": [{
    "index": 0,
    "message": {
      "role": "assistant",
      "content": "界面主要存在三处问题：1) 顶部导航栏与内容区未对齐，留白不均；2) 主按钮颜色对比度不足，弱化操作引导；3) 表单校验提示位置不统一。建议统一栅格、强化主按钮视觉权重，并将校验提示统一至输入框下方。"
    },
    "finish_reason": "stop"
  }],
  "usage": {
    "prompt_tokens": 512,
    "completion_tokens": 186,
    "total_tokens": 698
  }
}

Request Parameters

Parameter	Type	Required	Description
model	string	✓	Model ID; use glm-5.3-flash
messages	array	✓	List of chat messages; content may be a plain string or multimodal blocks containing image_url
thinking	object	—	Thinking config, {"type":"enabled"} (enabling only — disabling is not supported); clear_thinking:false is recommended
reasoning_effort	enum	—	Reasoning effort: low / high / max; max recommended
temperature	float	—	Sampling temperature, 0–2; 1 recommended
top_p	float	—	Nucleus sampling, 0–1; 0.95 recommended
max_tokens	int	—	Maximum output tokens
stream	bool	—	Whether to stream the response, default false; enabling tool_stream alongside it is recommended
tools	array	—	Tool definitions for function calling
tool_choice	string	—	Tool choice policy: auto / none / a named function