Wisdom Spectrum's "Flash Series" text, image understanding, image generation and video generation models are all free of charge!

About free model rate limiting

Total api call consumption $0 - $50/month (not included)

GLM-4-Flash: Concurrent 200

GLM-4V-Flash: concurrent 10

Cogview-3-Flash: Concurrent 5

CogVideoX-Flash: concurrent 3

GLM-4-Flash

present (sb for a job etc)

GLM-4-Flash Language Model is the first free API for Wisdom Spectrum AI, which is excellent in inference speed, long context processing, multi-language support, etc. It is suitable for a variety of application scenarios, such as smart Q&A, summary generation, and text data processing, etc. GLM-4-Flash combines the features of "speed" and "economy", and is very suitable for simple tasks that require fast response time. GLM-4-Flash combines "high speed" and "affordability", making it ideal for simple tasks that require fast response time.

Features

• Fast inference speed: GLM-4-Flash generates 72.14 tokens per second, which is equivalent to about 115 characters per second, and its speed is only second to GPT-3.5-turbo.This efficient inference speed enables the model to process large amounts of text data quickly, which is especially suitable for those high-frequency and simple task processing.
• Long Context Support: The model is equipped with 128K long context inference capability, which means that a single cue word can be processed with a text length of up to 300 pages. Such capability enables GLM-4-Flash to better understand and process long text content for scenarios that require in-depth context analysis.
• Strong multi-language support: GLM-4-Flash has strong multi-language support capability, which can support up to 26 languages. This provides multi-language interaction services for global users and broadens the application scope of the model.
• High concurrency: By adopting advanced technologies such as adaptive weight quantization and multiple parallelization methods, GLM-4-Flash achieves higher concurrency and throughput. This not only improves the processing capability of the model, but also significantly reduces the inference cost, making the model economical while maintaining high performance.

Scenario Applications

1. Text generation and editing

As an intelligent writing assistant, GLM-4-Flash can provide professional support for diverse writing scenarios. In the field of social media, it can assist in creating eye-catching Little Red Book tweets; in literary creation, it can provide suggestions on plot ideas and text embellishment; for commercial writing, it can help generate standardized reports, proposals and marketing copy. The model can not only grasp the writing characteristics of different genres, but also provide personalized modification suggestions according to the user's needs, making the writing process smoother and more efficient. Through AI assistance, it can significantly improve the quality and efficiency of writing.

clue

-上下文：我想推广公司的新产品。我的公司名为：智谱AI，新产品名为：ChatGLM大模型，是一款面向大众的AI产品。
-目标：帮我创建一条小红书平台的帖子，目的是吸引人们点击产品链接进行学习和体验。
-风格：参照Dyson等成功公司的宣传风格，它们在推广类似产品时的文案风格，同时结合小红书的文案风格。
-语调：说服性
-受众：AI产品在小红书上的主要受众是年轻人，活跃在互联网和AI领域。请针对这一群体在选择护发产品时的典型关注点来定制帖子。
-响应：保持小红书帖子简洁而深具影响力，注意要使用emoji表情，
**平台链接以markdown格式输出显示**：［智谱AI开放平台］（https://open.bigmodel.cn/console/trialcenter）。
**平台logo放在文案最下方**："D（https://s21.ax1x.com/2024/12/17/pALCRaT.png)*

Example of generated results

🌟【AI新潮流🌟】
亲爱的朋友们，你们是否还在为繁琐的日常事务烦恼？🤔
智谱AI全新力作——ChatGLM大模型，来啦！🎉
这款面向大众的AI产品，能帮你轻松应对各种挑战，让你的生活更加智能便捷！🎁
👉【一键解锁高效生活】
🔍精准搜索，告别无效信息
✏️智能写作，告别写作难题
📚知识问答，让你成为学霸
👨‍💻智能客服，解决你的燃眉之急
📢📢📢还在等什么？快来体验ChatGLM大模型，开启你的智能生活之旅吧！🌈
🔗点击链接，立即体验！🔗
[智谱AI开放平台](https://open.bigmodel.cn/console/trialcenter)
🌈🌈🌈
——智谱AI，与你一起，探索AI无限可能！
🔍🔍🔍
[D](https://s21.ax1x.com/2024/12/17/pALCRaT.png)

2. Large-model AI translation services

With its advanced deep learning architecture and powerful natural language processing capabilities, GLM-4-Flash is able to provide high-quality and efficient multilingual translation services. The model not only accurately grasps the semantic nuances of various languages, but also keenly understands the expression habits of different cultures, thus ensuring that the translation results are both accurate and authentic. In practical application, it supports instant mutual translation between dozens of mainstream languages, including Chinese, English, Japanese, Korean, etc., with fast response speed and high accuracy. Its excellent contextual understanding and terminology processing ability can meet the translation needs in different industries and scenarios, providing users with powerful support to break down language barriers and promote globalized communication and cooperation.

clue

翻译以下莎士比亚戏剧《罗密欧与朱丽叶》中的选段：
"To be, or not to be: that is the question:Whether 'tis nobler in the mind to suffer The slings and arrows of outrageous fortune,Or to take arms against a sea of troubles And by opposing end them."

Example of generated results

生存还是毁灭，这是一个值得考虑的问题：究竟是在心灵上忍受命运的箭矢和狂暴的打击，还是拿起武器对抗那片汹涌的烦恼之海，与之抗争到底。

3. High-efficiency legal document entity extraction, help the industry intelligent upgrading

The GLM-4-Flash model is equipped with efficient information extraction capability of legal documents and can accurately recognize the entity content in the documents. It effectively simplifies the process of document processing, greatly improves the efficiency of legal work, and provides a more convenient service experience for legal professionals.

clue

你现在是一个法律专家，请你对这篇判决书的内容进行分析。不要展现分析过程，直接按照下面的格式输出
## 判决书内容：
中华人民共和国最高人民法院
指定管辖决定书
（2017）最高法刑辖 19 号
关于被告单位北京盘古氏投资有限公司涉嫌骗取贷款、被告人吕涛等八人涉嫌骗取贷款、骗购外汇、非国家工作人员受贿、非法拘禁、故意毁坏会计凭证、会计账簿、财务会计报告等犯罪案件，本院经审查，依照《中华人民共和国刑事诉讼法》第二十六条的规定，决定如下：指定辽宁省大连市西岗区人民法院依照刑事第一审程序对该案进行审判。 二〇一七年三月十七日
## Output Format：
"犯罪客体"：""，
"犯罪主观要件-罪过形式"：｛
"故意"："",
"过失"：""
}
"犯罪主观要件-犯罪动机"："",
"犯罪主观要件-犯罪目的"："",
"犯罪客观要件-犯罪地点"："",
"犯罪客观要件-犯罪行为"："",
"犯罪客观要件-犯罪过程"：""，
"适用法条"："",
"判决结果时间"：""，
"判决刑期"：""，
"判决结果金额"：""

Example of generated results

＃ 输出格式
output_format =｛
"犯罪客体"："被告单位北京盘古氏投资有限公司及其相关人员"，
"犯罪主观要件-罪过形式"：｛
"故意"："骗取贷款、骗购外汇、非国家工作人员受贿、故意毁坏会计凭证、会计账簿、财务会计报告等行为"，
"过失"："无"
｝，
"犯罪主观要件-犯罪动机"："获取非法利益"，
"犯罪主观要件-犯罪目的"："非法占有"，
"犯罪客观要件-犯罪地点"：""，
"犯罪客观要件-犯罪行为"："骗取贷款、骗购外汇、非国家工作人员受贿、非法拘禁、故意毁坏会计凭证、会计账簿、财务会计报告等行为"，
"犯罪客观要件-犯罪过程"；""，
"适用法条"："《中华人民共和国刑事诉讼法》第二十六条"，
"判决结果时间"："二〇一七年三月十七日"，
"判决刑期"：""，
"判决结果金额"；
}

GLM-4V-Flash

present (sb for a job etc)

GLM-4V-Flash is the first completely free image understanding model introduced by Smart Spectrum. It is based on Wisdom Spectrum's 4V series of large multimodal models, with the ability to handle image-text interaction, and has demonstrated excellent performance in a number of tasks such as image recognition, image Q&A, and image reasoning. In enterprise application scenarios, GLM-4V-Flash has outstanding performance and can closely match the actual needs of different vertical fields, providing efficient large model image processing solutions for free.

Features

• Image Understanding and Processing: GLM-4V-Flash is able to automatically generate precise descriptive text for images, helping users to deeply understand image content. It can identify the main objects and scenes in an image, providing technical support for image classification and facilitating the organization and management of image content.
• Visual Reasoning and Interaction: The model has a strong visual reasoning capability to understand the relationship and interaction between the objects in the image. It supports users to obtain image-related information by asking questions, which enhances the interactivity and practical application value of the model.
• Multi-language support: GLM-4V-Flash supports 26 languages including Chinese, English, Japanese, Korean, and German, which greatly expands its application scope and enables it to serve users of different languages around the world.
• API call and Base64 encoding support: the model provides an API interface to facilitate users' image analysis. At the same time, it supports Base64 encoding for user uploaded images, which increases the flexibility and ease of use of the model.

basic competence

Image description generation

Prompt

Generate results

Provide a detailed description of the image, including major elements, scene, colors, etc.

This picture shows a cabin located in the forest. The cabin is built in a meadow surrounded by thick trees and bushes. The grass is in varying shades of green, and in some places the sunlight shines through the leaves, creating a contrast between light and dark. The cabin was constructed of wood and the roof was covered with gray shingles. The windows glowed yellow and there was a flight of steps leading up to the door. Some cut wood was stacked in front of the door. A meandering stream flows through the lower left of the image, adding a bit of life to this quiet forest. The overall color palette is predominantly cool, but the light inside the cabin brings warmth, giving it a serene and welcoming feel.

image classification

Prompt	Generate results
Look closely at the image and give the breed of dog in the image.	The dog in the picture would be a French Bulldog (French Bulldog).

visual inference

Prompt	Generate results
Based on the graph, speculate on the weather for that day	As you can see in the picture, the lady is wearing a blue raincoat and an umbrella. It can therefore be inferred that it may be raining or about to rain that day.

Visual Question and Answer (VQA)

Prompt	Generate results
I'll share with you a picture of a concert I saw recently.	The penguin is standing on stage, holding a microphone and confidently showing off its talent. As you can see from the photo, the penguin is wearing a formal suit, looking very elegant and professional. The stage background is uniquely designed to complement the penguin's image, giving it a visual treat.

Image Sentiment Analysis

Prompt	Generate results
Analyze and describe the primary emotion or mood conveyed by the image.	The expressions of the characters are full of joy and excitement: big smiles, wide eyes and hands raised as if in celebration convey a positive emotion.

Scenario Applications

Line Chart Analysis: A Concise Guide to Trend Forecasting

The Big Model GLM-4V-Flash has excellent charting and analyzing capabilities, and performs particularly well when dealing with line graphs. It can accurately identify and extract data inflection points, peaks and valleys in the charts, and use advanced algorithms to provide in-depth interpretation of data trends. By analyzing the changing patterns of historical data, the model can generate scientific predictions and present possible future directions in a clear way. This intelligent analysis capability not only significantly improves the efficiency of the decision-making process, but also provides managers with a more reliable basis for decision-making, helping companies grasp the first opportunity in the rapidly changing market environment.

Prompt

Generate results

Please help me analyze the trend of house prices in the picture and predict what the trend will be in the following year

The line graph comparing the trend of Chengdu's new and second-hand home sales volume in the past two years shows that from July to August 2022, the new home sales volume was on an upward trend; from September to December, the new home sales volume declined. From January to April 2023, the volume of new home sales continued to rise to 28,189 units; from May to June, the volume of new home sales declined again; in July, it rebounded again to about 23,790 units; the volume of new home sales fluctuated a lot in the following months, but generally stayed at a high level. For the new housing market trend in the coming year (2024), based on the trend shown in the chart, it is expected that the volume of new housing transactions will continue to maintain volatile growth, but the growth rate may slow down due to the impact of macroeconomic, policy regulation and other factors.

OCR Insurance Policy Information Extraction

GLM-4V-Flash is capable of efficiently extracting key information from insurance documents, including insurance amount, effective date, insurance terms and conditions. Its application not only dramatically improves the speed and accuracy of insurance business processing, but also saves insurance companies a lot of human resources. In the highly competitive insurance market, the system can help insurance companies improve service quality and optimize customer experience, thus solidifying their position in the industry and attracting more customers.

column1

column2

DEFINE ROLE AS "Invoice Recognition Expert". Knowledge Areas = [Invoice Recognition, Taxation, Financial Accounting] Skills = [text recognition, information extraction, format validation, data manipulation] Experience = "Senior" # defines the invoice field structure invoice_fields = { "Invoice Base Information": { "Invoice Type": {"description": "Type of invoice, e.g., VAT e-general invoice"}, "Invoice Code": {"description": "The 10-12 digit code in the upper left corner of the invoice"},. "Invoice Number": {"description": "The 8-digit number in the upper right corner of the invoice"},. “开票日期”: {“description”: “格式为YYYY年MM月DD日”}, "Check Digit": {"description": "The check digit in the upper right corner of the invoice."},... "Machine Number": {"description": "Invoice Machine Number"} }, "Purchaser Information": { "name": {"description": "Full name of the purchaser"}, "Taxpayer Identification Number": {"description": "Uniform Social Credit Code of the Purchaser"},. "Address Phone": {"description": "Address and contact phone number of the purchaser"},. "Account bank and account number": {"description": "Purchaser's account bank and account number"} }, "Seller information": { "name": {"description": "Full name of the seller"}, "Taxpayer Identification Number": {"description": "Uniform Social Credit Code of the seller"},. "Address Phone": {"description": "The address and contact number of the seller"},. "Account bank and account number": {"description": "Seller's account bank and account number"} }, "Product Information": { "Name of the goods or taxable labor service": {"description": "Name of the goods or service"},. "Specification": {"description": "The specification of the item."}, "units": {"description": "Units of Measurement"}, "Quantity": {"description": "Quantity of the item"},. "Unit price": {"description": "Unit price of the item"} }, "Amount Information": { "Amount": {"description": "Amount excluding tax"}, "Tax Rate": {"description": "Applicable Tax Rate"}, "Tax Amount": {"description": "Tax Amount"},. "Total_price_tax_in_capitalization": {"description": "Chinese capitalization of the total amount including tax"}, "Total_Price_Tax_Lower_Case": {"description": "A numerical representation of the total amount including tax."} }, "Biller Information": {"Payee": {"description": "Payee name "}. "Reviewer": {"description": "Name of the reviewer"}, "Biller": {"description": "Name of the biller"}, "Seller's stamp": {"description": "Information about the seller's stamp"} } } def extract_invoice_fields(image_content). Step1: Identify the invoice type Step2: Locate and extract fields by dimension Step3: Perform field validation and formatting Step4: Return structured data output = { category: {field: "" for field in fields.keys()} for category, fields in invoice_fields.items() } return outputMAIN PROCESS(image). user_input = read("""[image]"") return extract_invoice_fields(user_input) Strictly in json format, output the result of MAIN PROCESS execution, prohibit any additional interpretation:

{ "Invoice Base Information": { "Invoice Type": "Zhejiang VAT Electronic Ordinary Invoice". "Invoice code": "033002300611", "Invoice number": "94456370". "Invoicing Date": "06/26/2024", "Check digit": "78306379771971734013", "Machine number": "661201587811" }, "Purchaser Information": { "Name": "XXXXXXXX Company". "Taxpayer identification number": "913XXXXXXXXXX765N", "Address and phone number": "". "Account bank and account number": "" }, "Seller information": { "Name": "XXXXXX Ltd.". "Taxpayer Identification Number": "913XXXXXXXXXXXXXXQ9W". "Address Phone": "Zhejiang Province XXXXXXXXXXXXXX0520", "Account Bank and Account Number": "China Merchants Bank Limited, Nanjing Branch, Business Department 125916098210901" }, "Product Information": { "Name of goods or taxable services": "Transportation services* Passenger service charges". "Specification type": "", "Units": "", "Quantity": "1". "Unit price": "38.55" }, "Amount Information": { "Amount": "38.55". "Tax rate": "3%". "Amount of tax": "1.16". "Total price and tax_capitalization": "39.71". "Total_Price_Tax_Lower_Case": "39.71" }, "Billing Party Information": { "Payee": "Cai Jiao". "Reviewer": "Zheng Qi". "Biller": "Chen Le". "Seller's stamp": "" } }

Skin Type Picture Test Suggestions

GLM-4V-Flash's Beauty Consultation Assistant function recognizes skin problems in the picture and also provides a series of detailed skincare recommendations. It tailors the user's skin care regimen from a variety of perspectives, including cleansing, moisturizing, and sun protection, while emphasizing avoidance of inappropriate skin care behaviors and attention to lifestyle habits. This feature can provide users with advice on beauty matters.

Social Media Content Generation

Prompt

Generate results

Create an engaging Little Red Book (hiking) copy based on the content of the image

[Explore the unknown and challenge yourself] 🏞️ A walk through the landscape and through the clouds! 🌄 Sunshine on your body, breeze on your face, feel the charm of nature. 💪 Hiking is not only a way to exercise your body, it is also a cathartic experience. 📸 Take pictures of the beauty along the way and record every memorable moment. 💌 Share this joy and passion with your friends. 🔥 Come join us as we conquer the mountains and take in the views! # Hiking # Exploring Nature # Challenging Yourself

Educational applications

Prompt

Generate results

What physics is reflected in the diagram

The experimental phenomenon in the figure is dispersion of light. When white light passes through a prism, due to the different refractive indices of different colors of light, they are deflected at different angles, thus forming a spectrum of colors. This phenomenon is called dispersion of light. Historically, Newton used a prism to break down sunlight into the seven colors of the rainbow and proved that white light is a composite color, a discovery that had a profound impact on optics and the entire scientific community.

quality control

Prompt	Generate results
Identify how many bad fruits are in the picture	According to the part circled in red in the picture, there are three bad fruits in total. Bad fruits are usually fruits that have been damaged during picking, transportation or storage or whose quality has deteriorated due to diseases and other reasons. These bad fruits may affect the quality of other good fruits, so they need to be sorted and processed in time.

Product Description Generation

Prompt	Generate results
Generate a product title for the items in the image for the Taobao store!	Royal blue flowers] bone china teacup set, gold-rimmed flowers, delicate and elegant, showing noble temperament!

data annotation

Prompt	Generate results
accurately identify the type and color of the car in the image. and output it in json format	{"car_type": "Semi-truck", "color": "black"}

Cogview-3-Flash

present (sb for a job etc)

CogView-3-Flash is a free image generation model launched by Wisdom Spectrum AI, which can generate compliant images with higher aesthetic scores according to user commands.CogView-3-Flash is mainly used in the fields of art creation, design reference, game development, virtual reality, etc., and can help users to quickly realize the needs of conversion from text to image.

Features

• Multi-resolution support: The model supports a variety of resolutions, including 1024x1024, 768x1344, 864x1152, 1344x768, 1152x864, 1440x720, 720x1440, etc., which is able to satisfy the high standard of image quality required in the fields of professional design, advertising and publicity, and artistic creation.
• Creativity is rich and diverse: the model is able to generate images with rich creativity and imagination based on text descriptions inputted by the user, providing a wide range of sources of inspiration and creative possibilities for creative workers.
• Fast reasoning: the model has the ability to generate images in real time with fast response time, which can quickly meet the user's needs for image generation.

Scenario Applications

CogView-3-Flash is a highly efficient AI text-generated image model that can quickly generate high-quality images based on users' text descriptions. It is characterized by ultra-fast reasoning speed and accurate detail reproduction, and it takes only a few seconds on average to complete the generation of an image, so that the process of transforming ideas into visual works is more fluent and natural.

Prompt	Generate results
A young woman stood at a window, her hair falling softly over her shoulders, the The eyes were deep and dreamy.

1. CogView-3-Flash: AI assistant for generating PPT illustrations with one click

CogView-3-Flash can significantly improve the efficiency of PPT production, especially in the selection of background images. When we need a specific theme or style of PPT background image, we only need to describe the scene, style and color tone we want through text, and CogView-3-Flash can quickly generate a background image that meets the needs. Whether it is a business presentation, academic report or creative presentation, it can generate professional and beautiful background images according to the specific scene. This not only saves time in searching for materials, but also ensures the uniqueness of the background image, making the PPT visually appealing. With the assistance of AI, we can focus more on content creation and improve overall work efficiency.

Prompt

Generate results

Please generate a warm and educational background image that is suitable for use in a toddler drowning prevention PPT. The diagram should include a clear swimming pool or lake surrounded by safety features such as life preservers, life jackets, etc... as well as children playing happily with safety measures such as swimming rings. There should also be eye-catching safety signs in bright colors suitable for children's vision

2. Image backpropagation drawing: an intelligent closed loop from original image to AI reproduction

The powerful combination of Cogview-3-Flash and GLM-4v-flash brings users a brand new image creation experience. First, GLM-4v-flash, with its advanced image recognition technology, is able to accurately parse the content of the image you provide and generate a detailed and accurate description. Then, using this description, Cogview-3-flash is able to utilize its powerful image generation capabilities to create a new image that is similar in style and detail to the original. This process not only preserves the essence of the original image, but also allows you to incorporate your personalized ideas and preferences into the new image to get the visual work you have in mind.

CogVideoX-Flash

present (sb for a job etc)

CogVideoX-Flash is a free video generation model launched by Wisdom Spectrum AI, which inherits CogVideoX's self-developed end-to-end video comprehension model and has powerful video generation capability. CogVideoX-Flash not only supports text-generated video, which accurately generates corresponding dynamic video content based on the input text description, but also supports graph-generated video, which can deeply analyze the input image, extract key information, element characteristics and potential semantics in the image, and then extend and expand the video creatively based on it, which greatly enriches the source of materials and creative paths of video creation, and meets the diversified needs of users for video creation.

new feature

• Newly addedAI soundThe function makes your video not only visually stunning, but also brings unprecedented enjoyment in the sense of hearing.
• Tucson video supports up to 4K HD video, showing fine picture quality and enhancing the visual feast.
• Tucson video supports up to 10 seconds of video length to meet the needs of more scenarios.
• Tucson video supports up to 60fps, with dramatically improved smoothness, capturing every moment.

Generate Example

Vincennes Video Example

Toussaint Video Example

CogVideoX-Flash converts user-supplied still images into motion video. For best results, it is recommended to use the expression "subject (background) + motion description" for the cue.

Prompt	video
Female on blue background The camera rotates around her, and the light and colors in the background change as she moves, the Showcasing modern technology.
Make the picture move
The ostriches are talking about it.