At a Glance
| Platform | Price | Free Plan / Trial | Best For |
|---|---|---|---|
| Google Vertex AI AutoML (now under the Gemini Enterprise Agent Platform) | Usage-based, no flat platform fee — you pay separately for training hours, deployed endpoint-hours, and predictions (e.g. AutoML Vision training lists at $3.465/hour) | No AutoML-specific free tier; only Google Cloud’s general trial credit | Teams already on Google Cloud that want no-code tabular or image models without managing infrastructure |
| DataRobot | Not publicly listed — custom quote required | Free trial via sign-up (credit-based) | Enterprises wanting a governance-heavy, fully managed AutoML/MLOps platform and willing to negotiate a contract |
| Azure Machine Learning (AutoML) | No platform fee — you pay only for the underlying compute, billed by the second | No AutoML-specific free tier; general Azure trial ($200 credit / 30 days) | Teams in the Microsoft/Azure ecosystem wanting code-optional AutoML across tabular, vision, and NLP tasks |
Short Verdict
None of these three platforms charges a traditional “per-seat” software license for AutoML. Instead they split into two very different pricing philosophies: Google and Microsoft both meter the underlying compute you consume, while DataRobot keeps its pricing off the public web entirely and negotiates per account. If transparent, pay-as-you-go pricing matters most, Azure Machine Learning and Vertex AI AutoML are easier to budget for than DataRobot. If you need a single vendor to manage governance, monitoring, and explainability for you at enterprise scale, DataRobot’s opacity on pricing is the trade-off for that white-glove positioning.
What It Is: Google Vertex AI AutoML
Vertex AI is Google Cloud’s unified machine learning platform, and AutoML is its no-code/low-code layer for training tabular, image, and forecasting models without writing model architecture code. There’s an important naming wrinkle worth flagging for 2026 buyers: Google’s own Vertex AI documentation now states that the Vertex AI docs are “no longer updated” and that its services are “now part of” the Gemini Enterprise Agent Platform. The current pricing page for this stack doesn’t use the term “Vertex AI” at all — it’s framed entirely around “Gemini platform” consumption and “Gemini Enterprise Flexible Savings Plan” discount tiers, while noting that costs “remain the same as they are for the legacy AI Platform and AutoML products that Gemini platform supersedes.” In practice, the underlying AutoML training and serving mechanics haven’t changed — the branding and documentation home have.
On capability: AutoML for tabular data (classification, regression, forecasting) and AutoML Vision (image classification, object detection) remain available today. AutoML Text and AutoML Video, however, are confirmed deprecated — AutoML Text was deprecated on September 15, 2024 and shut down on June 15, 2025; AutoML Video was deprecated on July 31, 2024 and shut down on July 31, 2025. Google’s guidance is to replace both with Gemini prompting and tuning workflows (for example, fine-tuning Gemini on labeled text or video for classification, entity extraction, or sentiment tasks) rather than a dedicated AutoML text/video trainer. Several even older “Legacy AutoML” products (Legacy AutoML Natural Language, Video Intelligence, Vision, and Tables) were also retired between 2023 and mid-2024, superseded first by Vertex AI and now folded into the Gemini Enterprise Agent Platform.
What It Is: DataRobot
DataRobot is an enterprise automated machine learning and MLOps platform built around a model “leaderboard” — it trains and ranks many candidate models automatically, then adds governance, monitoring, and explainability tooling on top for production deployment. It’s positioned less as a developer tool and more as an enterprise platform purchase, which shows up clearly in its pricing: DataRobot’s own pricing page lists no dollar figures or tiers at all, only calls to action to “Request a Demo” or start a trial. Third-party enterprise software review site TrustRadius corroborates this, stating plainly that DataRobot “offers customized solutions to their customers, so to obtain an estimate or demo, contact DataRobot’s team.” A free trial using usage credits is available, but production pricing is quote-only and reportedly varies by consumption model (hosted cloud vs. credit-based) and by deployment scale.
What It Is: Azure Machine Learning (AutoML)
Here’s a naming correction worth making explicit, since it trips up a lot of comparisons: “Azure Machine Learning Studio” and “Azure Machine Learning studio” are not quite the same thing anymore. The original drag-and-drop “Machine Learning Studio (classic)” was a separate, older product that Microsoft stopped allowing new resources on starting December 2021 and fully retired on August 31, 2024, directing customers to migrate to Azure Machine Learning. Today, “Azure Machine Learning studio” refers to the web UI of the modern Azure Machine Learning service — and that’s where its current no-code AutoML experience lives. Within that current service, Automated ML supports classification, regression, and time-series forecasting through the no-code studio UI, plus computer vision (multi-class/multi-label image classification, object detection, instance segmentation) and NLP (text classification, named entity recognition) through the CLI/SDK.
On pricing, Microsoft’s official pricing page is unambiguous: “there is no additional charge to use Azure Machine Learning.” You pay only for the compute you provision — billed per second with no long-term commitment — plus any other Azure services you choose to layer on, such as Blob Storage, Key Vault, Container Registry, or Application Insights, which are billed separately from Azure Machine Learning itself.
Pricing Breakdown
| Platform | Training | Deployment / Hosting | Prediction |
|---|---|---|---|
| Vertex AI AutoML Vision | List price $3.465/hour (classification or object detection); $18.00/hour for edge/on-device training | $1.375/hour (classification) to $2.002/hour (object detection) while a model is deployed to an endpoint — billed even if idle | Batch prediction listed at $2.222/hour; 1-year and 3-year Flexible Savings Plan discounts reduce these rates (e.g. classification training drops to $3.1185/hr at 1 year, $2.772/hr at 3 years) |
| Vertex AI AutoML Tabular | $21.252 per node-hour (list) | Charged at standard custom-model inference rates (e.g. an n1-highmem-8 instance lists around $0.544/hour, plus any accelerator fees) | Same as deployment compute rate; Explainable AI is also charged at the inference rate |
| Vertex AI AutoML Forecasting | $21.252/hour (list) | N/A (prediction only) | Tiered per 1,000 forecasted data points, per month: $0.20 (first 1M), $0.10 (1M–50M), $0.02 (above 50M) |
| DataRobot | Not disclosed publicly | Not disclosed publicly | Not disclosed publicly — contact sales for a quote |
| Azure Machine Learning AutoML | No AutoML-specific fee — billed as standard compute by the second | No AutoML-specific fee — billed as standard compute | No AutoML-specific fee — billed as standard compute |
Pros
Vertex AI AutoML
- Transparent, published per-hour pricing for every stage (training, deployment, prediction)
- Deep integration with the rest of Google Cloud’s data and AI tooling
- Multi-year Flexible Savings Plan discounts for predictable workloads
DataRobot
- Strong model governance, monitoring, and explainability tooling aimed at regulated enterprises
- Automated model comparison (“leaderboard”) across many algorithm types in one run
- Free trial available before committing to a quote
Azure Machine Learning
- No platform fee at all — the simplest pricing model of the three
- No-code AutoML UI for tabular tasks plus SDK-based AutoML for vision and NLP
- Per-second compute billing with no long-term commitment required
Cons
Vertex AI AutoML
- Documentation and branding are mid-transition to the Gemini Enterprise Agent Platform, which can confuse teams searching for “Vertex AI” guidance
- AutoML Text and AutoML Video are gone — teams must re-architect around Gemini tuning instead
- A deployed endpoint bills continuously until explicitly undeployed, even with zero traffic
DataRobot
- No public pricing anywhere — budgeting requires a sales conversation before you know your cost
- Harder for small teams or individual developers to evaluate cost-effectiveness quickly
Azure Machine Learning
- No AutoML-specific free tier beyond the generic Azure trial credit
- No-code AutoML UI is strongest for tabular data; vision and NLP AutoML require the SDK/CLI, reducing the “no-code” claim for those task types
- Legacy naming overlap with the retired “Machine Learning Studio (classic)” still causes confusion in search results and older tutorials
Who It’s a Good Fit For / Who Should Look Elsewhere
Vertex AI AutoML fits teams already standardized on Google Cloud who want predictable, published unit pricing for tabular or vision models. Teams that relied on AutoML Text or AutoML Video should look elsewhere, or plan a migration to Gemini-based tuning instead.
DataRobot fits large, regulated enterprises that need governance and explainability baked in and have the procurement process to negotiate a custom contract. Smaller teams, startups, or anyone wanting to see a price before talking to a salesperson should look elsewhere.
Azure Machine Learning fits Microsoft-ecosystem teams that want the lowest-friction pricing model and are comfortable mixing a no-code UI (for tabular AutoML) with some SDK work (for vision/NLP AutoML). Teams wanting a single no-code UI that covers every data type equally well may find this split frustrating.
Bottom Line
There’s no single “cheapest” answer here because the pricing models aren’t comparable on the same axis: Vertex AI AutoML and Azure Machine Learning both meter compute transparently (Azure adds zero platform markup; Google publishes per-hour rates with volume discounts), while DataRobot simply doesn’t publish pricing and expects a sales conversation. For most teams building straightforward tabular or vision models inside an existing cloud, Vertex AI AutoML or Azure Machine Learning AutoML will be easier to budget for than DataRobot. DataRobot remains the right call primarily when governance and explainability tooling, not raw AutoML cost, are the deciding factor.
FAQ
Is “Vertex AI” still the correct name for Google’s AutoML platform?
Functionally, yes — the underlying training and serving APIs are the same — but Google’s current documentation says the Vertex AI docs are no longer updated and that these services are now part of the Gemini Enterprise Agent Platform. Expect more rebranding in docs and pricing pages going forward.
Can I still train a new AutoML Text or AutoML Video model on Google Cloud?
No. AutoML Text training ended with its June 15, 2025 shutdown, and AutoML Video training ended with its July 31, 2025 shutdown. Google directs new projects to Gemini prompting/tuning instead.
Does DataRobot publish any pricing at all?
No. Its official pricing page contains no dollar figures or tiers — only demo and trial sign-up links. A free trial is available, but production pricing requires contacting sales.
Is today’s “Azure Machine Learning studio” the same product as a few years ago?
Not quite. The old drag-and-drop “Machine Learning Studio (classic)” stopped accepting new resources in December 2021 and fully retired on August 31, 2024. “Azure Machine Learning studio” today is the web interface of the current Azure Machine Learning service, which includes the present-day AutoML feature.
Sources
- Vertex AI / Gemini platform pricing — Google Cloud: https://cloud.google.com/vertex-ai/pricing
- Vertex AI deprecations (AutoML Text, AutoML Video, Legacy AutoML) — Google Cloud Docs: https://docs.cloud.google.com/vertex-ai/docs/deprecations
- DataRobot Pricing page: https://www.datarobot.com/pricing/
- DataRobot Pricing Overview — TrustRadius (corroborating secondary source): https://www.trustradius.com/products/datarobot/pricing
- Azure Machine Learning pricing — Microsoft Azure: https://azure.microsoft.com/en-us/pricing/details/machine-learning/
- What is automated machine learning (AutoML)? — Microsoft Learn: https://learn.microsoft.com/en-us/azure/machine-learning/concept-automated-ml
- Machine Learning Studio (classic) retirement — Microsoft Learn Lifecycle: https://learn.microsoft.com/en-us/lifecycle/products/machine-learning-studio-classic
