Orderless Future

What is a gradeless future?

NoOrder Future is a platform for GPU leasing and AI applications, which brings together elastic computing power, model resources, application images, and container management under one website.

Users can either start ready-made applications such as ComfyUI directly, or create GPU instances to carry out tasks related to model training, inference, and development.

The main functions of Orderless Future

Computing power market

The computing power market is used to select GPU models, CPUs, memory, system disks, and images, and to create cloud instances based on the current prices of these resources.

• A variety of GPU models and configurations are available.
• Supports creating and shutting down instances on demand.
• The operating cost is calculated on a per-second basis.
• After shutting down, it can be started again and the data can continue to be used.

AI Application Market

App stores package common open-source tools, models, and dependencies into images; once a user selects a GPU, the application can be launched, thereby reducing the need for manual deployment steps.

• ComfyUI with commonly used nodes and models pre-installed
• Hunyuan Video, Flux, and Stable Diffusion workflows
• Model fine-tuning environments such as LLaMA-Factory
• Voice applications such as GPT-SoVITS and RVC
• Digital avatars, 3D and video creation tools

Model Square

The Model Plaza is used to discover the models supported by the platform as well as the entry points for invoking them; it is suitable for comparing different task types before selecting the appropriate application or computing resources.

Whether a model allows commercial use, redistribution, or the creation of specific content depends on the license provided by the model’s creator and the rules of the platform in question.

Container instance

The official documentation describes an instance as a cloud-based computer that can execute tasks when it is running, and while its operation is stopped it still retains the existing data.

Instances can still face risks due to accidental deletion, unpaid fees, platform rules, or storage issues; important models and works should not be stored solely in a single instance.

Public and private images

Public images provide an operating environment prepared by the platform, while private images allow users to save their own systems, dependencies, models, and working directories.

• Create instances quickly from public images
• Save the environment after debugging as a private image.
• Restore to working status using a private image.
• Reduce configuration time for teams or repetitive tasks

File transfer and remote access

The document offers various methods for uploading and downloading data, such as container cloud disks, file transfer plugins, cloud disks, FileZilla, and Jupyter.

• Connect to the cloud instance via SSH
• Use WebShell to perform operations in the browser terminal.
• Manage code and files through Jupyter Lab
• Configure the HTTP port for accessing the application interface

GPU monitoring and automatic shutdown

Users can monitor GPU usage and set up automatic shutdown for instances that are not in use, thereby preventing further consumption of the account balance after tasks are completed.

Automatic shutdown rules should first be tested on trial instances, and when training over extended periods, errors that may arise from intermittent data loading need to be taken into account.

How to use Unordered Future

For new users, it is recommended to first use instances with smaller specifications and for a shorter duration to test the image, network, and file transfer capabilities, before proceeding with actual training or generation tasks.

1. Register an account and complete the verification as required by the platform.
2. Top up a small amount of electricity or use the available campaign credit.
3. Select the task entry in the app market or computing power market.
4. Verify GPU, CPU, memory, storage, and pricing per unit
5. Create an instance and connect via web, SSH, or Jupyter.
6. Upload data and run small-scale testing tasks
7. Save the results, back up the files, and shut down the system in a timely manner.

Methods for selecting a GPU

• First, estimate the video memory usage based on the model parameters and precision.
• Training tasks take into account both GPU memory bandwidth and communication between multiple GPUs.
• When generating images and videos, it is necessary to check whether the application supports the graphics card architecture.
• Compare the per-second rate with the actual time taken to complete the task.
• Avoid focusing only on the graphics card name and ignoring the CPU, memory, and storage.

Supported platforms

• Web version or official online portal
• Linux

The client, region, language, and entry point may change with different versions; it is necessary to refer to the current product page before installation or payment.

APIs, SDKs, and open-source status

The official documentation provides APIs for GPU resource management, which can be used to query resources, create containers, obtain instance status, delete instances, and configure idle shutdown.

The platform also offers cloud-based workflows and guidelines for using certain models, which are useful for integrating tasks related to computing power creation and generation into automated processes.

Prices and packages

In the future, “electricity” will be used as the virtual balance for accounts, while GPU costs are deducted dynamically based on the model, configuration, and duration of use.

According to the official terms, electricity is considered a virtual product; once purchased, it generally cannot be returned or a refund issued. Therefore, it is necessary to confirm the available resources and the actual price per unit before making a purchase.

Which users are it suitable for

• AI painting users: Run ComfyUI and generation models in the cloud
• Video creators: Use video workflows such as Hunyuan and Wan.
• Model developer: Performs training, fine-tuning, inference, and testing
• Students and researchers: Rent GPUs as needed to conduct experiments
• Application author: Packages the image and distributes the application through the platform.

Product advantages

• Applications, models, and GPU computing power are concentrated on the same platform.
• Pre-built images lower the barriers to deploying open-source applications
• Supports SSH, WebShell, Jupyter, and port access.
• The document covers instances, images, files, and Linux operations.
• Provides APIs for GPU management and model invocation.

GitHub and the open-source status

The Unordered Future Platform is a closed-source commercial service; to date, no official repository containing its source code that can serve as a reliable reference for its identity has been found.

ComfyUI, models, and extensions available on the platform may come from different open-source projects; it is necessary to check each one regarding its license, commercial usage rights, and attribution requirements.

Usage restrictions and precautions

• GPU resources and unit prices vary depending on inventory and configuration.
• The electricity purchased usually does not qualify for a refund.
• Saving data when the device is shut down does not equate to a permanent backup.
• Third-party images and nodes may pose security and compatibility risks.
• When opening ports, authentication should be implemented to prevent sensitive services from being exposed.
• You need to verify the authorization for generating content and using training data on your own.

Basic information

Recommendation score

Recommendation score: 4.3/5.

Future Without Hierarchies is suitable for users who wish to quickly launch open-source AI applications or rent GPUs on demand; the application images and documentation available help to reduce the time required for environment setup.

Before use, it is important to compare the video memory capacity, unit price, storage options, and task execution time; moreover, automatic shutdown, data backup, and image security should be integrated into regular procedures.

Frequently Asked Questions

What platform is Unordered Future?

NoStep Future is a platform for GPU leasing and AI applications, offering cloud computing power, model and application images, container instances, as well as developer APIs.

Can ComfyUI be run directly with NoLevel Future?

Yes, app stores offer ComfyUI environments with commonly used nodes and models pre-installed, as well as related applications such as Hunyuan Video and Flux.

How will Noorder Future charge fees?

GPU instances are priced dynamically based on the selected model and configuration; the official website indicates that billing is done on a per-second basis and that electricity is consumed as well. The actual price is indicated on the page where the instance is created.

Will the data be lost after the instance is shut down?

The official documentation states that data can be retained after the instance is shut down and can be used again once it is restarted, but it is still necessary to create separate backups of important files.

Does Unorder Future provide APIs?

The official documentation covers querying GPU resources, creating containers, querying and deleting instances, automatic shutdown, cloud-based workflows, and model invocation.

Is Unordered Future an open-source platform?

The platform itself is a closed-source commercial service; many of the models and tools available in the app store come from open-source projects, but it is necessary to check each respective license.

Is it possible to get a refund after purchasing electricity?

The official purchase agreement defines electricity as a virtual product, stating that it is generally not refundable or exchangeable after purchase; it is advisable to conduct a small-scale test before making a payment.