GLM 5V Turbo

GLM-5V-Turbo

GLM-5V-Turbo is Zhipu's first multimodal Agent foundation model, deeply optimized for visual programming and complex task scenarios. It supports multimodal inputs including images, videos, text, and files, with enhanced visual understanding, long-horizon planning, and action execution capabilities. Compared with general-purpose multimodal models, it is better suited for integration into Agent workflows, completing the full closed loop of “environment perception → task planning → execution,” enabling multimodal capabilities to move from “being able to understand” to “being able to act.”

Context window131K

Provider智谱AI

Released2026/05/25

Playground

Pricing

The same model is available through multiple service channels — choose based on latency, reliability and cost.

Prices in $ / 1M tokens

To pick a channel, add a provider field to the request body, for example "provider": { "channel": "direct" }. Valid values are direct / stable / economical; omit it to use the default channel.

Direct

Direct upstream connection — best when you need native behavior and the full context window.

Input context	Input	Output	Cache read
≤ 32K	0.74/M	3.24/M	0.18/M
> 32K	1.03/M	3.82/M	0.26/M

Overview

Input

Text Image

Output

Text

GLM-5V-Turbo API: Zhipu's First Multimodal Agent Foundation Model

Compared with general-purpose multimodal models, GLM-5V-Turbo is better suited for integration into Agent workflows, completing the full loop of "environment perception → task planning → action execution" — taking multimodal capabilities from "being able to understand" to "being able to get things done."

SeaWhale AI provides GLM-5V-Turbo through an OpenAI-compatible interface, supporting multimodal inputs, tool calling, and streaming output.

Get API Key · Model ID: GLM-5V-Turbo

Why Choose GLM-5V-Turbo

Zhipu's first multimodal Agent foundation — not just understanding images, but making decisions and executing based on vision
Complete perception-planning-execution loop — integrated from environment understanding to action execution
Optimized for visual programming — writing code while looking at the UI, implementing frontend from design mockups
Multimodal inputs — unified processing of images, videos, text, and files
131K token context — ample multimodal context space

Core Capabilities

01 Visual Programming

GLM-5V-Turbo is deeply optimized for visual programming scenarios: writing frontend from design mockups, locating UI issues from screenshots, and generating data processing code from charts. This is the most direct engineering value of multimodal capabilities.

Design mockup to frontend code
Issue localization from UI screenshots
Code-based processing of charts and data

02 Environment Perception and Action Execution

As an Agent foundation model, its strength lies in converting visual understanding into concrete actions: recognizing UI elements, assessing the current state, and deciding the next step.

UI element recognition and localization
Current state assessment
Next-step action decision

03 Long-Horizon Planning

In visual tasks that require multiple steps, GLM-5V-Turbo maintains planning toward the overall goal rather than restarting judgment at each step.

Multi-step visual task planning
Intermediate state tracking
Goal consistency maintenance

04 Unified Multimodal Understanding

Images, videos, text, and files are processed uniformly in the same context, making it suitable for scenarios that require cross-modal associative reasoning.

Video content understanding and temporal reasoning
Mixed image-text understanding in documents
Cross-modal information association

Best Use Cases

Scenario	Description
Visual programming	Design mockup to frontend code, screenshot issue localization
UI automation	Vision-based UI operation and testing
Multimodal Agent	Agents that need to understand the environment before deciding and executing
Video content analysis	Temporal understanding and content extraction
Document understanding	Complete parsing of mixed image-text documents
Quality inspection and patrol inspection	Image-based judgment and subsequent action triggering

GLM-5V-Turbo vs. Other Multimodal Models

Capability	GLM-5V-Turbo	GLM-5 Turbo	Gemini 3.5 Flash
Model ID	`GLM-5V-Turbo`	`GLM-5-Turbo`	`gemini-3.5-flash`
Positioning	Multimodal Agent foundation	Fast tier for text-only agents	Efficient multimodal workhorse
Input modalities	Images, videos, text, files	Text	Text, images, videos, audio, PDF
Context window	131K token	131K token	105K token
Max output	32K token	131K token	64K token
Focus	Visual programming and action execution	Long execution chain stability	Coding and parallel agents

For specific pricing, refer to the real-time price card at the top of the page.

How to Use the GLM-5V-Turbo API

1. Create a SeaWhale AI API Key

Generate a key in the console and add credits.

2. Treat vision as part of the context

GLM-5V-Turbo's strength is not in simple image description, but in "what to do after seeing." Clearly stating the target action in the prompt yields the best results.

3. Call the API

curl -X POST https://api.atalk-ai.com/v1/chat/completions \
  -H 'Authorization: Bearer YOUR_API_KEY' \
  -H 'Content-Type: application/json' \
  -d '{
    "model": "GLM-5V-Turbo",
    "messages": [{
      "role": "user",
      "content": [
        {"type": "image_url", "image_url": {"url": "https://example.com/design.png"}},
        {"type": "text", "text": "Implement the corresponding Vue component based on this design mockup, paying attention to spacing and font sizes."}
      ]
    }],
    "stream": true
  }'

FAQ

How is it different from regular multimodal models?

Regular multimodal models aim to "understand," while GLM-5V-Turbo aims to "get things done after understanding" — it integrates visual understanding, long-horizon planning, and action execution into a closed loop, making it better suited for integration into Agent workflows.

What input modalities are supported?

Images, videos, text, and files. The output is text.

What exactly does the visual programming capability mean?

It includes converting design mockups to code, locating issues from UI screenshots, and code-based processing of chart data — directly converting visual information into actionable engineering output.

What are the context and output limits?

131,000-token context, with a maximum output of 32,000 tokens.

Is it suitable for UI automation?

Yes. It is optimized for the "environment perception → task planning → action execution" loop, making it a suitable choice for vision-based automation scenarios.

Can it process videos?

Yes. It supports video input and temporal understanding, suitable for content analysis scenarios.

Why Use GLM-5V-Turbo API with SeaWhale AI

Unified multimodal entry point — images, videos, documents, and text share a single set of APIs
OpenAI-compatible — integrate with existing code by changing just two lines
Complementary to text-only models — route visual tasks to 5V-Turbo and text tasks to the GLM-5 series
RMB settlement — domestic models are billed in RMB

API

API integration

Model IDUse this value as the model in inference requests

GLM-5V-Turbo

API KeyBearer token used to authenticate inference requests

Manage API keys

Base URLOpenAI compatible · /chat/completions

https://api.atalk-ai.com/v2/

provider OptionalSelects a service channel; omit it and the system picks the default

"provider": { "channel": "direct" }

GLM-5V-Turbo usage examples

SeaWhale AI is compatible with the OpenAI API protocol, so you can call it with the OpenAI SDK or plain HTTP requests. Streaming is enabled by default.

About the provider parameter (optional, a SeaWhale AI extension): most models are served over several channels that differ slightly in price and reliability. Add a provider field to the request body to pick one; omit it and the system selects the default channel — normal calls are unaffected.

provider is not part of the official OpenAI protocol — it is a SeaWhale AI extension that only takes effect on this platform. The OpenAI SDK allows custom fields like this to pass through; see the examples below.

Value	Channel	Best for
`direct`	Direct	The official upstream link, for native behavior and the full context window
`stable`	Preferred	Balanced availability and speed — a good default for production traffic
`economical`	Economy	Cost first, well suited to batch processing and price-sensitive workloads

Available channels and their prices are listed under "Pricing" above (channels vary by model). Additional notes:

Syntax: "provider": { "channel": "direct" }.
If the requested channel is not enabled for that model, the request automatically falls back to the default channel and returns normally — it does not error.
With the official SDKs: in Python pass it via extra_body; in Node.js put it directly on the request object and it passes through. In TypeScript projects, add a // @ts-expect-error line to skip the type check.

cURLPythonNode.js

curl https://api.atalk-ai.com/v2/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer <API_KEY>" \
  -d '{
    "model": "GLM-5V-Turbo",
    "messages": [
      {"role": "system", "content": "You are a helpful assistant."},
      {"role": "user", "content": "Hello!"}
    ],
    "provider": { "channel": "direct" },
    "stream": true
  }'
# provider is optional — remove this line to use the default channel

from openai import OpenAI

client = OpenAI(
    base_url="https://api.atalk-ai.com/v2",
    api_key="<API_KEY>",
)

stream = client.chat.completions.create(
    model="GLM-5V-Turbo",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Hello!"},
    ],
    stream=True,
    # Optional: pick a service channel; omit to use the default
    extra_body={"provider": {"channel": "direct"}},
)

for chunk in stream:
    if chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="", flush=True)

import OpenAI from 'openai'

const client = new OpenAI({
  baseURL: 'https://api.atalk-ai.com/v2',
  apiKey: '<API_KEY>',
})

const stream = await client.chat.completions.create({
  model: 'GLM-5V-Turbo',
  messages: [
    { role: 'system', content: 'You are a helpful assistant.' },
    { role: 'user', content: 'Hello!' },
  ],
  stream: true,
  // Optional: pick a service channel; omit to use the default
  // @ts-expect-error provider is a SeaWhale AI extension, not in the OpenAI SDK types
  provider: { channel: 'direct' },
})

for await (const chunk of stream) {
  process.stdout.write(chunk.choices[0]?.delta?.content ?? '')
}