Azure AI Studio pricing, which Microsoft now calls Microsoft Foundry pricing, confuses almost everyone at first. Azure OpenAI pricing adds even more strange words, like tokens, PTUs, deployments and commitment tiers. Then you wonder one simple thing: how much will this cost me? This guide explains Azure AI Studio pricing in plain words, so even a student in tenth grade can follow it.
Here is the short story of Azure AI Studio pricing. The platform itself is free to explore. You pay for the things you use inside it, and the biggest part of the bill is usually Azure OpenAI pricing, which means the cost of the AI models. In this guide you will learn how that cost works, see real examples with simple math, find seven hidden costs, and get ten ways to save money.
Azure AI Studio Pricing in 60 Seconds
- The platform is free. Microsoft says you can use and explore it without paying for the platform itself.
- You pay for usage. Each service inside it, such as models, search or agents, has its own price.
- Models are billed by tokens. A token is a small piece of a word. More text means more tokens and a higher bill.
- Big, steady workloads can reserve power. This is called Provisioned Throughput, or PTU.
- Use the Azure pricing calculator before you start, because prices change often.
| Question | Short answer |
|---|---|
| Is the platform free? | Yes, to explore |
| Do I pay for AI models? | Yes, per token |
| Do I need an Azure subscription? | Yes, to build agents and deploy models |
| Is there a way to get a fixed price? | Yes, PTU and commitment tiers |
| Can I limit my spending? | Yes, budgets and alerts |
This table sums up Azure AI Studio pricing. The next sections explain each line, starting with the names, because Azure AI Foundry pricing and Microsoft Foundry pricing mean the same thing.
What Is Azure AI Studio Today? Meet Azure AI Foundry
Many people still search for Azure AI Studio, but Microsoft has changed the name more than once. This is one reason pricing pages look confusing. You may find the same product under three names.
| Name | What it means |
|---|---|
| Azure AI Studio | The first name. A place to build and test AI apps. |
| Azure AI Foundry | The new name. Microsoft’s own post says Azure AI Studio was renamed to Azure AI Foundry. |
| Microsoft Foundry | The name on Microsoft’s current pricing page. It is the same platform, now with agents and more tools. |
So when you read “Azure AI Studio pricing,” think of Microsoft Foundry pricing. Azure AI Foundry pricing is the same thing too. The old web address still points to the new pricing page. If you see an older article, the idea is the same, but the prices and names may be old.
Azure AI Studio Pricing vs Azure OpenAI Pricing: What Is the Difference?
People often mix these two up. Here is a simple way to remember it. Think of a shopping mall. Azure AI Studio (Foundry) is the whole mall. Azure OpenAI is one big shop inside it. That shop sells OpenAI models such as the GPT family. Other shops in the mall sell other things, like models from other companies, search tools, safety tools and agent tools.
This matters for pricing. Azure OpenAI pricing is only one part of Azure AI Studio pricing, which covers the whole bill. The mall itself is free to walk around in, but every shop has its own price tags.
Is Azure AI Studio Free? The Truth About Azure AI Studio Pricing
Partly, yes. Microsoft’s pricing page says the Foundry platform is free to use and explore. You can look around without paying for the platform. However, Microsoft also says that each feature you use is billed at its normal rate. And before you can start building agents, you need an Azure subscription.
So the honest answer about Azure AI Studio pricing is this: looking is free, building costs money. If you only open the studio and read, you pay nothing. If you deploy a model and send it questions, you pay for those questions.
Is there free credit for testing Azure AI Studio pricing?
New Azure accounts have often received a starting credit. One recent pricing guide reports a $200 credit that lasts 30 days for new accounts, and no ongoing free tier after that. Offers change, so check the Azure free account page before you rely on this. Use the credit to test small ideas, and set a budget alert so a test does not turn into a surprise bill.
“Free platform” does not mean “free AI.” Azure AI Studio pricing mostly means paying for the models, and the models are the expensive part. Treat the platform like a free gym door and the models like a paid class inside.
How Azure OpenAI Pricing Works: Tokens Made Easy
Tokens are the main unit in Azure OpenAI pricing, and in Azure AI Studio pricing too. Almost every AI model on Azure is priced by tokens. A token is a small piece of text. It can be a whole short word, a part of a long word, or even a space or comma. A good rule of thumb is that 100 tokens equal about 75 English words. So a page with 500 words is roughly 670 tokens.
You pay for two kinds of tokens. Input tokens are the words you send to the model, such as your question and any files you attach. Output tokens are the words the model writes back. Output tokens cost more than input tokens, often several times more. This is because writing an answer takes more work than reading a question.
Cached input is cheaper
Some models also show a “cached input” price. If you send the same long text again and again, such as the same instructions at the start of every chat, the model can remember that part. Microsoft lists a lower price for cached input on many models. This can cut the cost of apps that repeat the same long instructions.
Microsoft also posted a notice about prompt cache writes. Starting November 1, 2026, if you ask for a 12-hour cache write, it is billed at the 12-hour cache-write rate. The default 30-minute option does not change. If you use caching, read this notice so your bill does not surprise you.
A simple way to picture tokens
Imagine a taxi. The meter runs for every kilometer you ride. Tokens are the kilometers. A short question is a short ride. A long document with a long answer is a long ride. The price per kilometer depends on the model you pick, just like a small car costs less than a big van.
The Three Ways to Pay in Azure AI Studio Pricing
Azure OpenAI pricing offers three main ways to pay, according to Microsoft’s pricing page. You can pick the one that matches how you use AI.
| Way to pay | How it works | Best for |
|---|---|---|
| Standard (pay as you go) | You pay per token, for input and output. No upfront promise. | Beginners, tests and small apps |
| Provisioned Throughput (PTU) | You pay an hourly price per PTU, even if you use less. Monthly and yearly reservations lower the cost. | Big, steady, always-on apps |
| Batch | Answers come back within 24 hours at a 50% discount on Global Standard pricing. | Jobs that can wait, like summarizing many files overnight |
Standard: like paying per ride
This is the default. You start with nothing and pay only for tokens used. If nobody uses your app this month, your model cost is close to zero. If lots of people use it, the cost goes up. It is the safest choice when you are learning or when you do not know how many people will use your app.
Provisioned Throughput: like renting a car
PTU means you rent a fixed amount of model power by the hour. You pay for it whether you use it or not. The benefit is steady speed and a predictable bill. Microsoft says the minimum is generally 15 PTUs for Global and Data Zone deployments and 50 for Regional ones, with 25 for some smaller Regional models. One guide estimates the starting cost at about $2,448 per month, so this option is for serious, steady workloads and not for beginners.
Batch: like sending a parcel by slow mail
If you do not need an answer right now, Batch is a great deal. You send a big pile of work, and the results come back within 24 hours at half the price of Global Standard. It fits tasks like summarizing thousands of documents, tagging old data or writing product descriptions in bulk.
Global, Data Zone and Regional: where your data is processed
Each deployment also has a location type. Global lets Microsoft run your request in many places. Data Zone keeps processing inside a zone, such as the EU or the US. Regional keeps it in one region, and Microsoft lists up to 27 regions. Stricter location rules can change the price and the minimum PTUs. If your company has privacy rules, ask which type you need before you plan your budget. The location type can change your Azure AI cost, so it is a key part of Azure AI Studio pricing.
Azure OpenAI Pricing: Example Model Prices
Azure OpenAI pricing is shown per one million tokens. The table below uses Global Standard prices reported in a May 2026 pricing guide. Microsoft’s own page shows exact prices for your region and currency, and it says its numbers are estimates. So treat these as examples, and confirm the live Azure AI Studio pricing in the Azure pricing calculator.
| Model | Input (per 1M tokens) | Output (per 1M tokens) | Good for |
|---|---|---|---|
| GPT-4o mini | $0.15 | $0.60 | Chatbots and high-volume, simple tasks |
| GPT-4.1 | $2.00 | $8.00 | Long documents (very large context) |
| GPT-4o | $2.50 | $10.00 | Smart answers, images and agents |
| Mistral Large 3 | $0.50 | $1.50 | Many languages and instruction tasks |
| Phi-4-mini | about $0.07 | about $0.23 | Sorting and pulling out facts (indicative) |
Microsoft’s page now lists newer GPT-5 family models, the o-series reasoning models, audio, image and video models. Each has its own price columns, and some have Batch, Flex or Priority options. The examples above are older, popular models that are easy to compare. Always pick your exact model in the calculator.
Other things that have prices
Models are not the only price tags on the page. Microsoft also lists prices for embeddings (per 1,000 tokens), image models such as DALL-E 3 (per 100 images), fine-tuning for training and hosting, and video models such as Sora (per second). Voice and audio models are priced per million tokens, and some translation and speech tools are priced per hour.
Real Azure AI Studio Pricing Examples With Simple Math
Numbers make this clear. These three examples use the example prices above. The formula is easy: million input tokens times the input price, plus million output tokens times the output price.
Example 1: A small help chatbot
Your chatbot answers 10,000 chats a month. Each chat uses about 800 input tokens and 400 output tokens. That adds up to 8 million input tokens and 4 million output tokens. Look at how much the model choice matters.
The same chatbot costs $3.60 on the small model and $60 on the bigger one. Both can work well for a simple help desk. This is why choosing the right size of model is the biggest money saver.
Example 2: Summarizing 2,000 documents
Each document has about 6,000 input tokens, and each summary has about 500 output tokens. The total is 12 million input and 1 million output tokens.
| Model | Cost on Standard | If Batch gives 50% off |
|---|---|---|
| GPT-4o mini | $2.40 | about $1.20 |
| GPT-4.1 | $32.00 | about $16.00 |
| GPT-4o | $40.00 | about $20.00 |
The Batch column assumes the model supports Batch and you can wait up to 24 hours. It is an example of the discount, not a quote.
Example 3: A company search tool (RAG)
RAG means the AI first finds pieces of your company files and then writes an answer from them. Say 50,000 questions a month, each with about 3,000 input tokens (the question plus the found text) and 300 output tokens. That is 150 million input and 15 million output tokens.
| Part of the bill | GPT-4o mini | GPT-4o |
|---|---|---|
| Model tokens | $31.50 | $525.00 |
| Search service (Azure AI Search Basic, reported) | about $75 | about $75 |
| Rough total | about $106 | about $600 |
Notice two things. First, the model choice again changes the total a lot. Second, the search service is a real cost that many beginners forget. This brings us to the hidden costs.
7 Hidden Costs in Azure AI Studio Pricing That Surprise Beginners
Most surprise bills do not come from the model. They come from the small things around it. Azure AI Studio pricing has many parts, so check this list before you start.
1Search for your own data
If you let the AI search your files, you usually need Azure AI Search. One guide reports about $75 a month for the Basic level and about $250 a month for Standard S1. Microsoft also says vector search needs an embedding model, and embeddings are billed by use. Search runs all month, even when nobody is asking questions.
2Hosting a fine-tuned model
Fine-tuning teaches a model your style or data. The training job itself can be cheap. One guide says a small job can cost about $5 to $10. The surprise is hosting. A fine-tuned model must stay deployed, and Microsoft lists hosting by the hour for many models. One guide estimates about $50 to $70 a day for a fine-tuned GPT-4o endpoint. Delete the endpoint when you finish testing.
3Tool sessions and file storage for agents
Agent tools have their own meters. Microsoft lists Code Interpreter as billed per session, and each session is active for one hour by default. File Search is billed per GB of vector storage per day, with 1 GB free, plus a charge per 1,000 tool calls. Small amounts add up when many users run agents.
4Web grounding
If your AI uses Grounding with Bing Search to get fresh facts from the web, Microsoft says this is billed separately from the model tokens. Do not assume the web lookup is included in the token price.
5Supporting services
Microsoft’s own post says Azure AI hubs, Azure OpenAI and projects are billed by compute hours and tokens. It also lists supporting services such as Key Vault, Storage, Application Insights and Event Grid, which charge by transactions and data volume. One guide estimates blob storage at about $0.02 per GB each month and monitoring at roughly $35 to $50 a month per environment. Each item is small, but they stack.
6Network and data transfer
Microsoft notes that a hub’s network is public by default. If you secure it with private networking, you may pay extra data transfer costs. This is good for safety, so plan for it instead of avoiding it.
7Output tokens and long chats
Output tokens cost more, and long chats send the whole history back to the model each time. A chat that runs for fifty messages can cost far more than fifty short, separate questions. Set a limit on answer length, and clear old history when you can.
Leaving a test running. A deployed fine-tuned model, an idle search service or a reserved PTU keeps charging while you sleep. Before you log off, delete or stop anything you do not need.
10 Ways to Save Money on Azure AI Studio Pricing
Good news: Azure AI Studio pricing gives you many ways to save. Use these ten steps to lower your Azure AI cost and your Azure OpenAI pricing bill.
- Start with the smallest model that works. Test GPT-4o mini or another small model first. Move up only if the answers are not good enough.
- Use Batch for work that can wait. You get 50% off Global Standard pricing when you can wait up to 24 hours.
- Keep prompts short. Every extra word is a token you pay for. Remove long, repeated text.
- Use cached input. Put the same long instructions at the start, so repeated text can use the cheaper cached price.
- Limit the answer length. Ask for “three bullets” or set a maximum output size.
- Set budgets and alerts. Create a budget in Azure Cost Management so you get a warning before the bill grows.
- Set tokens-per-minute limits. A cap on each deployment stops runaway use from a bug or a busy user.
- Use separate resource groups and tags. Microsoft suggests dedicated resource groups or subscriptions for hubs so you can see exactly who spent what.
- Consider a commitment tier for steady use. It is a discounted alternative to pay as you go. You pay a fixed fee for a month, and usage above the quota is charged at the overage rate. Purchases are non-refundable.
- Use PTU only when usage is high and steady. If your traffic goes up and down, pay as you go is often cheaper.
Microsoft also lists an AI Flex Plan. It is one commitment across eligible Microsoft AI services, including Foundry, for a one-year or three-year term. It is only available on the Microsoft Customer Agreement for Enterprise, and it does not change list prices.
How to Check Your Azure AI Studio Price Before You Spend
You do not have to guess. Follow these steps to get a real estimate before you build.
- Open the Azure pricing calculator. Microsoft says its prices are estimates, and the calculator gives you the numbers for your own region and currency.
- Add the services you plan to use. Include the model, search, storage and any agent tools.
- Estimate your tokens. Use the rule that 100 tokens equal about 75 words, and guess how many chats or files you will handle each month.
- Look at the Example Scenarios tab. Microsoft says the calculator can show upfront monthly costs for AI Foundry scenarios.
- Create a budget. Add a budget and an alert in Azure Cost Management before you deploy anything.
- Review after one week. Compare the real bill to your estimate and adjust the model or limits.
Which Azure AI Studio Pricing Option Fits You?
Each type of user has a different best option. Find your row in the table.
| If you are… | Best choice | Why |
|---|---|---|
| A student or beginner | Standard | No promise, and tiny costs for small tests |
| A small business with a chatbot | Standard, small model | Low cost and easy to start |
| A team that processes many files | Batch | Half price if you can wait |
| An app with steady, heavy traffic | PTU or commitment tier | Predictable cost and steady speed |
| A big company with a Microsoft agreement | AI Flex Plan | One commitment across AI services |
6 Mistakes That Raise Your Azure AI Cost
These mistakes push up Azure OpenAI pricing and your total Azure AI Studio pricing bill.
- Picking the biggest model first. Most simple tasks do not need it.
- Forgetting the extras. Search, storage and monitoring are part of the real cost.
- Leaving things running. Delete test deployments, endpoints and unused services.
- Reading old pricing articles. Names and prices change, so use the live calculator.
- Buying PTU too early. Reserve power only after you know your steady usage.
- Skipping alerts. A budget alert takes five minutes and can save you hundreds.
Azure AI Studio Pricing: Questions People Ask
Is Azure AI Studio free?
Azure AI Studio pricing starts with a free platform. The platform is free to use and explore, according to Microsoft’s pricing page. You pay for the features you use inside it, such as AI models, search and agents. You also need an Azure subscription before you can build agents.
How much does Azure AI Studio cost per month?
There is no fixed monthly fee for the platform. Azure AI Studio pricing depends on what you use. Microsoft says hubs, Azure OpenAI and projects are billed by compute hours and tokens. As an example from this guide, a small chatbot with 10,000 chats a month costs about $3.60 on GPT-4o mini and $60 on GPT-4o in model tokens.
Is Azure AI Studio the same as Azure AI Foundry?
Yes. Azure AI Foundry pricing and Azure AI Studio pricing describe the same platform. Microsoft states that Azure AI Studio was renamed to Azure AI Foundry. Microsoft’s current pricing page uses the name Microsoft Foundry. It is the same platform with newer names and more tools.
What is the difference between Azure AI Studio and Azure OpenAI?
Azure AI Studio (now Foundry) is the whole platform for building AI apps. Azure OpenAI is a service inside it that gives you OpenAI models such as GPT. The platform is free to explore, while Azure OpenAI is billed by use.
How much does Azure OpenAI cost?
Azure OpenAI pricing depends on the model and how much you use it. Models are billed per token. As an example, one May 2026 guide lists GPT-4o mini at $0.15 for input and $0.60 for output per million tokens, and GPT-4o at $2.50 and $10.00. Check the Azure pricing calculator for current prices in your region.
How is Azure OpenAI billed?
Azure AI Studio pricing has three main ways to pay. Standard bills per token with no promise. Provisioned Throughput (PTU) bills an hourly price per unit of reserved power. Batch gives a 50% discount on Global Standard pricing when you can wait up to 24 hours for results.
What is a token in Azure OpenAI?
A token is a small piece of text, like a short word or part of a long word. About 100 tokens equal 75 English words. You pay for input tokens (what you send) and output tokens (what the model writes), and output usually costs more.
What is a PTU in Azure OpenAI?
PTU stands for Provisioned Throughput Unit. It reserves model power for you at an hourly price, whether you use it or not. Microsoft lists minimums of generally 15 PTUs for Global and Data Zone and 50 for Regional. It fits large, steady apps.
Is there a free tier for Azure OpenAI?
There is no ongoing free tier for models, according to one recent pricing guide. New Azure accounts have often received a starting credit, reported as $200 for 30 days. Check the Azure free account page, because offers change.
Is Azure OpenAI cheaper than the OpenAI API?
It depends on the model, region and deployment type. List token prices for the same model are often close, so compare both official price pages for your exact model. Many teams choose Azure for its security, location and billing controls, and not only for price.
How can I check my Azure AI cost before I start?
Use the Azure pricing calculator. Add your model, estimate your tokens and include search and storage. Microsoft says the calculator can show monthly costs for example scenarios, and that its page prices are estimates, not quotes.
How do I limit my Azure AI spending?
Create a budget and alerts in Azure Cost Management, set tokens-per-minute limits on each deployment, use tags and separate resource groups to track costs, and delete test resources when you finish.
What is commitment tier pricing?
It is a discounted alternative to pay as you go. You commit to a fixed fee for a calendar month and get a set amount of usage. Usage above that is charged at the overage price. Purchases are non-refundable, and you can turn off auto-renewal before the month ends.
Do I pay for Azure AI Studio hubs and projects?
Microsoft says hubs, Azure OpenAI and projects are billed by compute hours and tokens used, and not by a fixed monthly fee. Supporting services such as storage, Key Vault, Application Insights and Event Grid add small charges based on use.
Final Thoughts on Azure AI Studio Pricing
Azure AI Studio pricing is easier than it first looks. The platform is free to explore. You pay for tokens, for any reserved power and for the extra services around your app. Pick a small model, keep prompts short, use Batch for work that can wait, and set a budget alert on day one.
Most important, estimate before you build. Open the pricing calculator, try your own numbers and check the bill after a week. Prices and names change, so treat every article, including this one, as a guide and confirm the final Azure AI Studio pricing on Microsoft’s official page.
Ready to estimate your Azure AI cost?
Pick one idea, such as a help chatbot. Write down how many chats you expect each month, then enter it into the Azure pricing calculator. You will have a real Azure AI Studio pricing number in ten minutes.