Skip to main content
Here are some frequently asked questions about Novita AI. Before contacting our support team, please check the FAQs below to help you quickly find solutions.

​
GPU Instance

​
1. How to check the price of the GPU instances?

You can check the price of GPU instances and their configurations (container disk, volume disk, network volume, etc.) on the Pricing Page.

​
2. When does the billing for GPU instance start?

Billing starts when the instance status changes to “Pulling” status.

​
3. Introduction of container disk, volume disk, and network volume.

  • Container Disk
    • Does not support dynamic expansion, can only specify capacity when creating an instance;
    • Mount directory: / (cannot be customized);
    • Data will be saved when saving the image;
    • Supports 60GB free quota, charges will apply for the excess part, for details refer to: Billing Instructions
  • Volume Disk
    • Supports dynamic expansion;
    • Default mount directory: /workspace (customizable);
    • Data will not be saved when saving the image;
    • Read and write speed is the same as the container disk;
    • Volume Disk capacity requires additional charges, for details refer to: Billing Instructions.
  • Network Volume
    • Supports dynamic expansion;
    • Default mount directory: /network (customizable);
    • Network volume has an independent lifecycle, unrelated to the instance, even if the instance is deleted, the network volume data still exists;
    • Overall read and write speed is slower than the container disk or volume disk (depending on specific usage);
    • Network volume capacity requires additional charges, for details refer to: Billing Instructions

​
4. Why can’t the instance be restarted after it stops?

This applies to pay-as-you-go instances only. Monthly subscription instances have pre-reserved resources and will not encounter this issue.
After a pay-as-you-go instance stops, its underlying resources are released back to the resource pool. If those resources have since been claimed by other instances, the instance cannot be restarted. In this case, it is recommended to first save the image based on the target instance, and then create a new instance based on the saved image.
After saving the instance image, the data on the container disk will be saved with the image, but the data on the volume disk will not. It is recommended to use the network volume for data with high persistence requirements.

​
5. How to handle abnormal instance status?

First, try to troubleshoot the problem through the “System Logs” and “Instance Logs” of the instance. If the problem cannot be resolved, you can contact us.

​
6. No instance specifications with a specified CUDA version.

CUDA versions are backward compatible. For example, if your service relies on CUDA version 12.1, you can choose an instance specification with a CUDA version greater than or equal to 12.1.

​
7. What is the maximum CUDA version supported by the platform?

You can check the allowed CUDA versions in the “Filter” module at the bottom right corner of the Explore.

​
8. How to diagnose the “Save Image” failure?

First, try to troubleshoot the problem through the logs of the “Save Image” task. If you are saving the image to a private repository address, please check whether your Container Registry Auth Configuration is correct. If the problem cannot be resolved, you can contact us.

​
9. Can dedicated IP be supported?

Yes. Currently, this capability is not open to the public. If you have such requirements, please contact us.

​
10. How to check the GPU usage of the instance?

Due to the PID isolation of Docker containers, the nvidia-smi command cannot be used to view the process. You can install the py3nvml library and use the shell command to check the GPU usage: 
# Install the py3nvml library.
$ pip install py3nvml
# Check the GPU usage.
$ py3smi
Fri Sep 20 12:17:39 2024
+-----------------------------------------------------------------------------+
| NVIDIA-SMI                        Driver Version: 550.54.14                 |
+---------------------------------+---------------------+---------------------+
| GPU Fan  Temp Perf Pwr:Usage/Cap|        Memory-Usage | GPU-Util Compute M. |
+=================================+=====================+=====================+
|   5 35%   28C    8   11W / 450W |   353MiB / 24564MiB |       0%    Default |
+---------------------------------+---------------------+---------------------+

+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
| GPU        Owner      PID      Uptime  Process Name                   Usage |
+=============================================================================+
+-----------------------------------------------------------------------------+

​
11. Can a single instance mount multiple network volumes?

Yes. Both the console and API support mounting multiple network volumes on a single instance.
Attaching multiple network volumes will trigger an automatic instance restart. Please save your work before proceeding.

​
Serverless GPUs

​
1. What is the difference between Serverless GPU and GPU Instance?

Serverless GPU is designed for deploying model inference endpoints without managing infrastructure. You define an endpoint, and the platform handles scaling, scheduling, and resource allocation automatically. GPU Instance gives you a persistent, interactive VM-style environment where you manage the runtime directly. Choose Serverless GPU for production inference workloads; choose GPU Instance for development, training, or workflows requiring persistent storage and interactive access.

​
2. How is Serverless GPU billed?

Serverless GPU is billed based on actual execution time — you are only charged when your endpoint is actively processing requests. There are no charges during idle periods. Pricing is calculated per second of GPU time consumed. For detailed rates, refer to the Serverless GPU pricing page.

​
3. What should I do if my Serverless GPU endpoint returns a timeout error?

Timeout errors typically occur when:
  • The cold-start time exceeds the request timeout threshold (e.g., the container image is large and takes time to pull).
  • The model inference time exceeds the configured request timeout.
Recommended actions:
  1. Use the image prewarm feature to reduce cold-start latency.
  2. Increase the request timeout setting for your endpoint if your workload requires longer processing times.
  3. If the issue persists, contact us with your endpoint ID and error details.

​
Payments

​
1. How can I avoid top-up failures?

Top-up failures are generally caused by two main reasons:
  • Rejection from the card issuer. This may occur for the following reasons. Please check or contact your card issuer for details:
    • The corresponding payment channel is not activated.
    • The credit card has expired or been frozen.
    • The credit card balance is insufficient.
    • The card number is incorrect.
    • The security code is incorrect.
  • Risk control measures from the payment channel. Please check and make any necessary adjustments:
    • The device ID is associated with a high number of cards.
    • The number of cards declined using this email address is very high.
    • The time since this card was first seen on the Stripe network with this device ID is very short.
    • The authorization rate associated with this email address is very low.
    • The name on the email address does not match the name on the card.

​
2. What payment methods are supported?

We accept credit card payments via Stripe. PayPal payments are also supported and are processed manually within 7 business days. Cryptocurrency payments are not currently supported.

​
3. How do I access my invoices?

Receipt links are accessible for 30 days following a successful payment. If your link has expired, please contact support with your order number to request a reissue.

​
4. Can I get a refund after topping up my account?

Refunds are generally not available. In exceptional circumstances (e.g., complete service unavailability), refund requests may be submitted to our support team and are typically processed between the 10th and 15th of each month.

​
Account

​
1. How do I delete my account?

Please provide your Novita account ID or registered email address to our support team. We will process your account deletion request within 14 business days.

​
2. Can I use the Coding Plan with a negative account balance?

No. Your account balance must remain positive to access the Coding Plan, even if you have an active subscription.

​
3. Why did my $1 credit voucher disappear?

Our system automatically revokes vouchers when suspicious registration activity is detected (e.g., malicious sign-ups or multiple accounts from the same device). If you believe this was done in error, please contact our support team for a reissuance review.

​
4. How do I change my account email address?

Contact our support team with your current registered email address and the new email address you wish to use. Our team will assist you with the update.

​
Enterprise

​
1. How do I inquire about enterprise requirements (e.g., H100 clusters)?

Please provide your company name, business use case, required GPU quantity, and contact information. Our account management team will reach out to you directly. You can also book a call with our sales team.