[ad_1]
Launched in 2021, Google Vertex AI is Google Cloud’s integrated machine learning platform that combines all of Google Cloud’s AI capabilities into a single user interface and SDK. Because so many components are rolled into one service, there is some confusion about what exactly is offered for free with Vertex AI and what exactly needs to be paid for. In this article, we’ll provide an overview of Vertex AI’s pricing and free credits so you understand what to expect when using this platform.
See more: What is Google Vertex AI?
Is Google Vertex AI free?
While not completely free, Vertex AI does offer:
Vertex AI offers $300 free credits and over 20 free products
- $300 in free credits to use on paid Vertex AI products and services for the first 12 months
- Free access to over 20 AI products including AutoML Vision, Video Intelligence, Natural Language and more
These free credits and products make it easy to get started with Vertex AI and experiment with different machine learning capabilities at no cost. The $300 credit is good for most early prototyping and proof-of-concept workloads.
Costs are similar to older services, with some exceptions
With multiple Google Cloud AI services consolidated into Vertex AI, the pricing model retains most of the previous pricing for those existing services, such as AutoML and AI Platform. So if you’re already familiar with Google Cloud AI pricing, switching to Vertex AI probably won’t drastically impact your costs.
There are two exceptions:
- Co-hosting and elevations: Vertex AI enables “co-hosting” of ML models on the same resources and charges container usage in 30-second increments instead of a minimum usage time. Both policies offer more flexible and potentially cheaper hosting options.
- Generative AI: During the current Preview period, Generative AI, including text, image, audio, and video generation, is currently discounted at $1 per 1,000 characters of input text and $1 per 1,000 characters of output text. Generating text via Vertex AI is therefore very affordable temporarily, but will probably be more expensive later.
Free model hosting and pay-as-you-go forecasting
Vertex AI does not charge a minimum fee for deploying and hosting machine learning models. You only pay per second for what you use. This applies both during training and when used to make predictions. For example, if your endpoint is called 60,000 times per month, you will be charged that month’s prediction fee based on hourly rates per node.
Some other important notes about model hosting pricing:
- When you deploy models, you are billed per node hour for each node in your endpoint. Any computing power associated with the endpoint is charged at the training rate.
- Auto scaling is available to scale nodes up/down based on traffic. More nodes means higher costs.
- Inference units determine the computing power each node supports. More powerful units cost more per hour.
So while hosting costs can add up at scale, you have the flexibility to control expenses through nodes, auto-scaling policies, and inference units.
Also read: Is Undress AI real?
What are Vertex AI Pipelines and TensorBoard Pricing?
In addition to models and predictions, Vertex AI includes two other products with their own prices:
Vertex AI Pipelines costs based on pipeline routes. There are currently no fees during the Preview period. Once launched, the cost is $0.03 per pipeline run. Complex pipelines spanning multiple steps can become more expensive to operate over time.
Vertex AI TensorBoard, currently in Preview, will charge a flat monthly fee of $300 per unique active user at launch, regardless of platform usage. Even if you only use it one day per month, you will be charged this monthly fee.
So while pipelines and access to TensorBoard are free today, keep in mind that costs will pile up in the future as they leave Preview.
Generative AI discounts apply during preview
As previously mentioned, Vertex AI is currently offering discounted pricing for Generative AI during the launch period. Some important notes:
- You are charged based on every 1,000 characters of input text in the model and every 1,000 characters of text generated in the response.
- The charge is $1 per 1,000 characters in either direction, input or output.
- However, currently this fee is reduced by 100% during the Free Access Preview Period.
Once you leave Preview, expect text generation pricing to follow the $1 per 1000 character discount model. This may still offer savings compared to other cloud providers.
View the billing details to understand the full costs
As shown above, different Vertex AI components have their own pricing models and discounts during Preview periods. To get a full picture of what you may be charged during your use:
- Consult the billing reports in the Google Cloud Console for your project
- View detailed breakdowns showing all Vertex AI SKUs with usage and costs
- Set billing alerts to be notified of unusually high charges
Constantly monitoring your billing dashboards is essential to avoid surprise invoices as costs increase due to production workloads.
Conclusion: weigh free credits versus extra costs
In summary, while Vertex AI consolidates many AI products and services into one platform, parts of them still charge standard Google Cloud fees, similar to the previous standalone services. Take advantage of free credits, sample discounts and pay-per-use models to minimize costs wherever possible. But keep a close eye on your billing console to understand when your usage exceeds free levels and be proactive in managing long-term costs. Please contact the Vertex AI team if you have any questions about pricing as you explore the possibilities of this new offering.
🌟 Do you have burning questions about a “Google Vertex AI”? Do you need some extra help with AI tools or something else?
💡 Feel free to email Pradip Maheshwari, our expert at OpenAIMaster. Send your questions to support@openaimaster.com and Pradip Maheshwari will be happy to help you!