Learn the main differences between the OVHcloud managed tools (AI Notebooks, AI Training, and AI Deploy) for building your machine learning projects.
Overview
OVHcloud AI tools work inside our Public Cloud ecosystem. Once logged in to your Public Cloud project, you have access to three AI tools:
- AI Notebooks
- AI Training
- AI Deploy
Each of them is an autonomous tool and is billed per minute, but they have shared elements, such as security tokens, etc.
All AI Tools have access to OVHcloud Object Storage as a central point, allowing you to work with your data securely and easily.
We schematize our AI offers this way:
AI Tools comparison
The tables below show the features and abilities of each of the three AI tools:
- AI Notebooks: Managed Jupyter or VSCode notebook, linked to compute resources (CPU or GPU) and storage.
- AI Training: Automate your trainings, deploy your own Docker container, linked to compute resources (CPU or GPU), and attached storage.
- AI Deploy: Deploy your own AI application and models with high availability via Docker containers or select one from our catalog.
Please take note that whatever other difference may be between them, all three tools are billed per minute.
| AI Features | AI Notebooks | AI Training | AI Deploy |
|---|---|---|---|
| Managed Jupyter Notebook with preinstalled libraries | X | ||
| Managed VSCode Notebook with preinstalled libraries | X | ||
| Automate Training phases (timeout, grid computing, ...) | X | ||
| Bring your own Docker image (with notebooks inside, models, etc.) and connect to compute and storage resources | X | X | |
| Deploy your managed applications and models with high-availability | X | ||
| Deploy AI applications from a catalog made with partners | X |
| Management | AI Notebooks | AI Training | AI Deploy |
|---|---|---|---|
| Control panel | X | X | X |
| API | X | X | X |
| CLI | X | X | X |
| Resources | AI Notebooks | AI Training | AI Deploy |
|---|---|---|---|
| CPU resources | X | X | X |
| CPU flavors available | ai1-1-CPU | ai1-1-CPU | ai1-1-CPU |
| GPU resources | X | X | X |
| GPU flavors available |
L4-1-GPU L40s-1-GPU |
L4-1-GPU L40s-1-GPU |
L4-1-GPU L40s-1-GPU |
| RAM Memory | X | X | X |
| Dask computing | X | ||
| Cold-patch resources (adjust GPU/CPU, attached storage) | X | ||
| Local SSD-NVMe storage | X | X | X |
| Local SSD-NVMe storage size | depends on selected flavors | depends on selected flavors | depends on selected flavors |
| Ability to attach a remote storage (Object Storage) | X | X | X |
| Pull Storage Sync (from object storage) | X | X | X |
| Push Storage sync (to object storage) | X | X | |
| Maximum runtime | unlimited | unlimited | unlimited |
| Ability to add a runtime timeout (via CLI) | X |
| Security | AI Notebooks | AI Training | AI Deploy |
|---|---|---|---|
| ACL via Openstack user | X | X | X |
| ACL via tokens | X | X | X |
| Backups and resiliency | |||
| Automatic backup | |||
| Backup when you stop | X | ||
| High availability | X (when replicas > 1) |
||
| Auto scaling | Planned |
| Advanced features and parameters | AI Notebooks | AI Training | AI Deploy |
|---|---|---|---|
| Labels | X | X | X |
| Start / Stop / Resume | X | X | |
| Easily import data from Git/* via UI | X | ||
| Unchanging endpoint URL | X |
Go further
For more information and tutorials, please see our other AI & Machine Learning support guides or explore the guides for other OVHcloud products and services.
If you need training or technical assistance to implement our solutions, contact your sales representative or click on this link to get a quote and ask our Professional Services experts for a custom analysis of your project.