Qikuai AI

What is Qikai AI?

Qikuai AI is a one-stop multi-modal creation and AI interface platform operated by Xiamen Yunkede Network Technology Co., Ltd. It brings together functions such as e-commerce video production, product photography, voiceover services, interview analysis, and the use of models from various suppliers under a single account and balance system.

The platform is intended for both content teams and developers: the former can use web templates to create marketing materials, while the latter can generate keys and integrate text, image, video, or voice models into their own systems. Different creation tools and models are charged separately; it is not possible to apply the price or capabilities of one model to all services.

Main functions

Replication of product videos and AI-driven commercial photography

• Product videos are created by replicating the shot structure and presentation style of existing videos; users select a template and upload product-related materials to generate similar marketing videos.
• AI-based commercial photography allows for the combination of product images, characters, scenes, props, audio, and scripts, making it suitable for developing video sequences from individual product images.
• The templates cover e-commerce categories such as clothing and bags, jewelry and accessories, beauty and skincare products, food and beverages, products for mothers and babies as well as pets, 3C home appliances, home furnishings, sports and outdoor items, and toys.
• The project process includes script generation, selection of segmented videos, video review, creation of the final video, and its download; the team can make further modifications or regenerate the video from the project list.

Creative freedom on the Quick Whiteboard

• QuickWhiteboard is designed for video projects where one does not want to be constrained by fixed templates; it allows users to organize images, videos, characters, scenes, and audio materials within a blank project structure.
• The user first determines the aspect ratio of the video, the key features of the product, and the order of the shots; thereafter, the content is generated segment by segment with subsequent filtering. This approach is suitable for product introductions, social media videos, and creative clips.
• Increased freedom also means a need for more precise management of scripts and assets; for complex projects, it is advisable to use low-cost segments first to verify consistency in characters, products, and style.

AI-powered commercial photography for jewelry display

• The jewelry wearing function takes in images of jewelry products and creates visuals showing how the jewelry would look on models, thereby reducing the need for separate photography sessions for each product.
• The results can be used for selecting options, testing compositions, and creating marketing drafts, but they do not guarantee that the size, material texture, and skin tone will be exactly identical to those of the actual product.
• Before uploading real people or images containing faces, permission for the processing of such portraits and personal information must be obtained; on official e-commerce pages, it is also necessary to verify the authenticity of the products as well as comply with the platform’s rules.

AI voice synthesis

• Mirror Dance Voice converts text into speech, and can be used to create voiceovers for product videos, live presentations, narrations, and various versions of advertisements.
• The user enters the dialogue and selects a voice to generate audio, which is then added to the video project to match the rhythm of the footage.
• The voice synthesis is based on third-party voice technologies, and the quality of the sound, the language used, the duration, availability, and pricing may vary depending on the product configuration; do not impersonate real persons or clone voices without permission.

Analysis of intonation-based interview responses

• This product accepts recorded audio or video interview materials, analyzes their content to produce background investigation reports; the platform also offers interfaces for submitting tasks, checking results, and viewing a list of projects.
• It is suitable for organizing interview notes, extracting key information, and drafting initial versions; the outputs of this model should not be used as the sole basis for making decisions regarding employment, dismissal, or credit assessment.
• Before processing information related to candidates, referees, or employees, legitimate authorization must be obtained, and authorized personnel must verify the facts, context, identities, and sensitive data.

Model Plaza and Input/Output

Model Plaza organizes models based on labels such as Anthropic, DeepSeek, Google, MiniMax, Moonshot, OpenAI, ByteDance, Zhipu, Alibaba, etc., and allows users to assign custom aliases to these models. The labels indicate how the platform categorizes models; they should not be construed as indicating that the manufacturers behind these models are directly selling them, providing guarantees for them, or having any specific partnership with the platform.

Web page creation tutorial

1. Register and log in to your account, then check the material requirements, sample results, and cost associated with the desired creation.
2. Based on the task, choose between reproducing product videos, using AI for commercial photography, working with a whiteboard, showcasing jewelry, providing voiceovers, or conducting interviews – this way you can avoid repeated attempts using inappropriate methods.
3. Prepare the images of the products for which usage rights are desired, as well as any material related to people, videos, audio files, and text; remove any unnecessary personal information before uploading them.
4. Select a template, aspect ratio, character, scene, or sound to create a short clip or sample first, in order to check the appearance of the product, the consistency of the characters, as well as the alignment between lip movements and subtitles.
5. Regenerate or replace the incomplete segments, verify all sections, combine them to form a complete project, and download a backup.
6. Before publication, manually check the price, efficacy, trademarks, portraits, copyrights, and AI-generated identifiers; retain the licenses for the original materials as well as records of the project.

API integration process

1. Register a corporate or developer account, select the desired products or models, and ensure that the account balance is sufficient to cover the testing calls.
2. A 64-bit key can be created for API key management, and an IP allowlist can be configured; keys must be stored only in the backend and in the security configuration system.
3. Read the model details and development documentation to understand the request format, model aliases, input types, asynchronous query methods, unit prices, and error codes.
4. Send requests in JSON format. Text models can use formats compatible with OpenAI, Anthropic, or Gemini; the specific format to be used depends on the endpoint types specified for that model.
5. First, use a small number of requests in the testing environment to verify various scenarios such as authentication, streaming responses, file size limits, timeouts, insufficient balance, and failed content filtering.
6. Before going live, enable the allowlist, rate limiting, retry mechanisms, log data masking, and cost alerts; also verify the charging amounts and results by comparing task logs with usage logs.

API restrictions and security recommendations

The development documentation provides example requests in Python, Java, PHP, and other languages; however, no officially released SDK is available at present. The code examples are intended to illustrate how requests should be made, and they do not represent open-source source code for the platform’s products.

Prices and payment methods

The top-up page offers payment options such as WeChat, Alipay, and corporate transfers; the available fixed amounts are 200, 500, 1000, 2000, 5000, and 10000 yuan. There is no information available regarding any free credits granted upon registration or permanent free usage limits; it is necessary to check the account balance and the price of each individual product before making a purchase.

The price of a model varies depending on its alias, context, input and output parameters, caching settings, image specifications, or video length. The table is provided solely to illustrate the current billing structure; the actual cost shall be based on the details of the model, the creation page, and the actual charge records.

Refunds, copyrights, and commercial use

• Platform services are virtual online products that require payment in advance before the service is provided; once activated, they cannot be transferred or refunded, unless there are serious defects that render them completely unusable, as required by law, or as determined by the platform.
• The intellectual property rights for the content uploaded, synthesized, and saved by users belong to the users or the original owners, provided that the users have legitimate authorization to use the materials, images, trademarks, and sounds involved.
• Content that has not been published on the platform and is only saved locally or stored temporarily on the server – according to the terms of the agreement, the platform and its affiliated parties do not acquire any rights to use such content; it is deleted automatically once the temporary storage period expires, with the specific duration not yet made public.
• The platform will add legal-compliant explicit or implicit identifiers to the generated synthetic content. For certain paid services, it is possible to request that no explicit identifiers be displayed, but it is still required to make lawful declarations and include relevant identifiers when such content is shared with others.
• Before using it for commercial purposes, it is necessary to check the terms of the upstream models, the licenses for the materials used, as well as the rules regarding advertising and the platforms on which content is published. Making a payment or obtaining the generated results does not automatically grant rights to third parties.

Privacy and data storage

• The platform may handle accounts, login and search queries, IP addresses, device types, access behavior, order payments, customer service communications, as well as photos and videos uploaded by users.
• Personal information collected and generated during operations within the country is stored within that country, and is retained for as long as it is necessary to provide the services; it is deleted or anonymized after the necessary period has elapsed, except in cases where legal retention is required.
• In the face-effect scenario, the original image is deleted immediately after the effect is generated, while the resulting image is stored in the cloud for free for 15 days. This time limit does not apply to other videos, audios, or API projects.
• Some functions may be handled by third-party service providers; do not submit passwords, identification documents, trade secrets, or unauthorized facial, voice, and interview data before invoking the models.
• Users can request access to, correction of, deletion of, or copying of their personal information; they can also contact customer service to delete their account. Account deletion is irreversible – the relevant data is deleted or anonymized, unless otherwise provided by law.

Platforms, open source, and usage boundaries

The Model Plaza includes labels and aliases for different manufacturers; whether each upstream model makes its weights available, permits commercial use, or offers an original SDK must be determined on a case-by-case basis for that particular model. The fact that the platform provides compatible interfaces does not mean that these models or the Qikuai AI products themselves are open source.

Suitable for users and scenarios

• E-commerce operations: Use product images to quickly create short videos, draft commercial content, and various versions of marketing materials.
• Jewelry merchants: Generate visuals of how the jewelry will look when worn, for use in image selection and creative testing; thereafter, human verification is carried out to confirm the authenticity of the products.
• Content team: Integrates scripting, voiceover, segment creation, and final video management within the same project workflow.
• Developer: Test various text, image, video, and audio models through compatible interfaces, and view usage metrics in a unified manner.
• Human Resources team: Conducts interviews under legitimate authorization and manual review; it cannot make high-impact decisions automatically.

Advantages and limitations

Main advantages

• Put e-commerce video templates, creative editing tools, jewelry display options, voice-over services, and model interfaces in the same platform to reduce the need to switch between multiple accounts.
• The model details include information on input and output, billing units, compatible endpoints, and code examples, which helps developers estimate the effort required for integration.
• It provides API keys, allowlists, task logs, usage logs, and balance records, making it suitable for transitioning from testing to production environments.

Usage restrictions

• The details of videos, characters, and products generated may be distorted; it is necessary to verify the actual product, the text descriptions, its functions, and the brand elements before carrying out any marketing efforts.
• The model’s price and availability status can change; therefore, it is not possible to base long-term budgeting or to lock in a model alias solely on the results of a single test.
• The fact that a platform aggregates multiple models does not mean that all of these models have identical functions, contexts, content policies, and service stability.
• To date, no unified free quota, native mobile applications, official SDKs, or open-source code for the product have been identified.

Frequently Asked Questions

Can Qikuai AI be used for free?

At present, there is no verified permanent free quota or free bonus offered upon registration. Web page creation and model interfaces are charged according to the specific product or model; it is necessary to check the balance as well as the details of the pages and models before using them.

How to create videos with Qikai AI?

Choose between video replication of a product, AI-generated commercial footage, or a blank template; upload the product and relevant materials, then configure scripts, characters, scenes, and sounds to generate the content segment by segment. By reviewing each segment before combining them for download, the cost associated with repeatedly generating the entire video can be reduced.

What API formats does Qikai AI offer?

Model details may list OpenAI, Anthropic, or Gemini-compatible endpoints; the specific models that are supported vary. The open platform uses Bearer tokens for authentication and provides product APIs as well as documentation for task querying.

Can a refund be obtained after making a top-up?

In principle, virtual services cannot be transferred or refunded once they have been activated. Exceptions may apply in cases where serious defects render the service unusable, as required by laws and regulations, or with the approval of the platform; it is advisable to conduct a trial use with a small amount of funds before making a full payment.

Can the images and videos generated by Qikuai AI be used for commercial purposes?

Users or the original rights holders retain the right to upload and create synthesized content, but commercial use is still subject to the permissions related to the materials, individuals, trademarks, sounds, underlying models, and distribution channels. The generated results do not automatically eliminate any risks related to third-party rights.

Does Qikuai AI save the uploaded facial images?

In the face-effect scenario, the original image is deleted immediately after the effect is generated; the resulting image is stored in the cloud for 15 days. The retention period for other types of creative content and API data may vary, and such data should be managed in accordance with the rules applicable to each product, with regular backups being made.

Is Qikai AI an open-source project?

No. At present, no product source code or open-source license has been found; development documentation and example requests do not constitute open source. Moreover, the availability of certain upstream models in the Model Plaza is independent of the platform itself.

Summary

Qikuai AI is suitable for teams that need to create e-commerce videos in bulk or to utilize multiple types of AI models in a unified manner. In practice, it is advisable to first test the functionality and interfaces using small-scale tasks, and then establish a comprehensive process that covers keys, balances, material licensing, generated identifiers, personal information, and manual review.