Qu Shi AI

What is QuShi AI?

QuChi AI is an artificial intelligence platform owned by Runjian Co., Ltd., designed for software projects and the implementation of AI solutions in enterprises.

Platform connection requirement analysis, interaction prototypes, custom development, large model APIs, and private deployment.

Core services

AI requirement analysis

Users can describe their ideas in natural language, and the AI will ask follow-up questions regarding the scenario, the user, and the functions.

• It is suitable for turning vague ideas into clearer software requirements.
• Business objectives, user roles, core processes, and permission relationships can be added.
• When the requirements are complex, it is necessary to clarify the data sources, exception handling, and approval steps.
• The requirement documents generated by AI can only serve as a starting point for discussions; they cannot replace formal reviews.
• When it comes to budget, timeline, and compliance, further confirmation is required by the project team.

Generation of AI prototype interfaces

Once the requirements are confirmed, the platform can generate a browsable, interactive prototype to facilitate quick communication regarding the solution.

• You can get started by entering the product scenario, target users, and core features.
• Suitable for websites, administrative backends, data dashboards, and industry application prototypes.
• If not satisfied, you can return to the requirements analysis phase to make further modifications.
• The prototype is used to test the information architecture and interactions; it is not software that can be launched directly.
• Brand visual identity, accessibility, performance, and accurate data still require further design and development.

Custom software development

Qu Shi AI offers customized services ranging from prototyping to delivery, with human consultants available to address ongoing requirements.

• Once the requirements are confirmed, the process of evaluation, quotation, development, and testing begins.
• According to the website’s claims, implementation can take as little as a week; however, the actual time required depends on the scope of the project.
• Delivery methods can include source code, cloud deployment, or on-premises deployment.
• Ongoing maintenance is part of the scope of services, and the details should be specified in the contract.
• Acceptance criteria, intellectual property rights, and change costs need to be defined prior to development.

Large model API services

The platform aggregates multiple mainstream models, and reduces the costs associated with working with various suppliers through a unified console and interfaces.

• The public model series includes DeepSeek, Qwen, GLM, Kimi, and MiniMax, among others.
• The types of capabilities include text dialogue, reasoning, embedding, multimodality, and image generation.
• The official website claims to support over twenty models, but the actual list of available models is subject to what is displayed on the Model Plaza.
• Pay-as-you-go is suitable for testing different models before deciding on a production solution.
• The model version, context length, and throttling rules may change depending on the supply.

Unified interface and migration

The official documentation states that the interface is compatible with the OpenAI format, which facilitates the migration of existing applications after adjusting their configurations.

• Developers need to create API keys in the console and store them properly.
• Before changing the service address and model name, it is necessary to read the corresponding model parameter instructions.
• Not all models support the same parameters, tool calls, or multimodal inputs.
• The production environment should have settings for timeouts, retries, concurrency, and rate limits.
• After migration, it is necessary to retest the output structure, costs, latency, and security policies.

Model selection recommendations

Tutorial for Creating a Curved-Angle AI Prototype

1. Log in to the platform and go to the section for software customization or prototype creation.
2. Describe the business scenario, target users, devices used, and core objectives.
3. List the main pages, user roles, data fields, and key operations.
4. Answer the AI’s follow-up questions by providing additional information on permissions, procedures, exceptions, and approval requirements.
5. After generating the prototype, check each page for navigation, forms, status, and operation feedback.
6. Return the functions that were modified inaccurately or omitted during the requirement phase.
7. Invite real users to try out the prototype and record their feedback.
8. After confirming the scope, contact the consultant to get an estimate of the cost and delivery timeline.
9. Before signing the contract, clarify the terms regarding source code, deployment, operation and maintenance, acceptance, and intellectual property rights.

Guide to Integrating the Curved Scissors AI API

1. Register and complete the account verification required by the platform.
2. Enter the model plaza to compare model capabilities, context, and real-time prices.
3. Create an API key in the console; do not include that key in public code.
4. Configure the service address, authentication header, and model name according to the official documentation.
5. Test text generation or other target capabilities using minimal requests.
6. Track the billing performance for incoming, outgoing, cached, and failed requests.
7. Add timeout, retry, rate limiting, log masking, and key rotation.
8. Use real business samples to conduct tests on quality, security, latency, and cost.
9. After passing the acceptance test, gradually increase the production volume.

Prices and billing methods

As of August 30, 2026, the platform employs a model that combines pay-per-use API pricing with project-based service quotes.

Models may be added or removed from the lineup, promotions and unit prices can change; the final details are subject to the pricing page before making a request, the billing information in the console, and the contract terms.

How to estimate API costs

• Count the numbers of inputs, outputs, cache hits, and multimodal requests separately.
• Use real business samples, rather than relying solely on estimates based on short, one-time conversations.
• Long contexts significantly increase the amount of input, so irrelevant historical messages should be restricted.
• Inference models may generate more outputs and result in longer waiting times.
• For batch tasks, costs related to retries, failures, and traffic peaks also need to be taken into account.
• After going live, billing monitoring should be carried out by separating it according to models, applications, and departments.

Private deployment

The privatization option is intended for enterprises that require exclusive use of resources, local storage of data, and deep system integration.

• You can choose a local data center, dedicated resources, or a hybrid deployment based on your business needs.
• It is necessary to evaluate model authorization, GPU resources, concurrency, and disaster recovery capabilities.
• The storage strategies for business data, logs, and vector databases should be designed separately.
• Upgrades, monitoring, security patches, and fault response should be included in the scope of services.
• Privatization is usually more costly and is suitable for organizations with specific compliance or scale requirements.

Key points for the acceptance of corporate projects

• Define project boundaries with a clear list of features and items that are not included.
• Prepare normal, abnormal, and permission test cases for critical processes.
• Clarify the scope of the source code, as well as the ownership of third-party components and models.
• Test performance, concurrency, backup, recovery, and security auditing.
• Confirm the timelines for migration, training, operation and maintenance, as well as response times.
• Acceptance should be based on milestones, to avoid judging the level of completion solely on the appearance of the prototype.

Platform, SDK, and open-source status

• Web: Offers prototypes, a model gallery, a console, and a documentation center.
• HTTP API: It enables developers to integrate model capabilities into existing applications.
• SDK: The official documentation center provides information related to the SDK; the specific language used is as indicated in the current documentation.
• Desktop and mobile clients: No official download page available for ordinary users.
• GitHub: No official public repository that corresponds explicitly to the current business platform was found.
• Open source: The QuShi AI platform does not come with an open source license made public.

The delivery of the source code for custom projects is stipulated in the contract; it does not imply that the QuShu AI platform or its underlying services are open source.

Which users are it suitable for

• Startup team: Quickly turn product ideas into presentable prototypes.
• Product Manager: Organize requirements and accelerate cross-departmental communication.
• Development team: Unified access and comparison of multiple large models.
• Small and medium-sized enterprises: Custom software development services, from identifying needs to having the software ready for use.
• Large organizations: Adopt private deployment to meet data and integration requirements.
• Industry solution providers: Develop applications by integrating models, data, and existing systems.

Usage restrictions and precautions

• An AI prototype is not a production system; it cannot replace engineering development and testing.
• The fastest delivery time stated on the official website is just a promotional figure; the actual project timeline is determined by the contract.
• A unified interface cannot eliminate the differences in parameters and outputs among different models.
• API keys should be stored on the server, with permissions, quotas, and a rotation mechanism in place.
• A privacy, confidentiality, and compliance assessment must be completed before uploading business data.
• Private deployment still requires ongoing operation and maintenance, security updates, and model governance.
• When comparing prices, quality, latency, stability, and the cost of manual verification should all be taken into account.

Frequently Asked Questions

What services does QuShi AI mainly provide?

It offers AI requirement analysis, interactive prototypes, large-model APIs, software customization, and enterprise-specific deployment.

Can QuShi AI directly create software that can be launched online?

AI can quickly generate interactive prototypes; however, before going live, it is still necessary to conduct requirement reviews, carry out development, testing, deployment, and acceptance processes.

Which large models is QuShi AI capable of supporting?

The public model series includes DeepSeek, Qwen, GLM, Kimi, and MiniMax, etc.; the current list is subject to Model Plaza.

How is the Curved Scissors AI API priced?

The API is billed on a pay-as-you-go basis; the public starting price is 0.28 yuan per million tokens, with the specific input and output costs varying depending on the model.

Does Curved AI support private deployment?

Yes, exclusive resource allocation, localized data storage, and system integration solutions are available; the price requires separate evaluation.

Is Curved AI compatible with the OpenAI interface?

The official documentation states that it is compatible with the OpenAI interface format, but the differences in model parameters and capabilities still need to be tested one by one.

Is QuShi AI an open-source platform?

It is not an open-source platform, and the delivery of the source code for custom projects does not equate to the making of the platform’s source code available to everyone.