Azure OpenAI Service pricing overview
To help customers in the journey, we offer pricing and cost management solutions to meet your needs. including:
- Standard (On-Demand): Pay-as-you-go for input and output tokens.
- Provisioned (PTUs): Allocate throughput with predictable costs, with monthly and annual reservations available to reduce overall spend.
- Batch API: Language models are also now available in the Batch API for global deployments and three regions, that returns completions within 24 hours for a 50% discount on Global Standard Pricing.
- Global Deployment – Global SKU
- Data Zone Deployment – Geographic based (EU or US)
- Regional Deployment – Local Region (up to 27 regions)
Explore pricing options
Apply filters to customise pricing options to your needs.
Prices are estimates only and are not intended as actual price quotes. Actual pricing may vary depending on the type of agreement entered with Microsoft, date of purchase, and the currency exchange rate. Prices are calculated based on US dollars and converted using London closing spot rates that are captured in the two business days prior to the last business day of the previous month end. If the two business days prior to the end of the month autumn on a bank holiday in major markets, the rate setting day is generally the day immediately preceding the two business days. This rate applies to all transactions during the forthcoming month. Sign in to the Azure pricing calculator to see pricing based on your current programme/offer with Microsoft. Contact an Azure sales specialist for more information on pricing or to request a price quote. See frequently asked questions about Azure pricing.
US government entities are eligible to purchase Azure Government services from a licensing solution provider with no upfront financial commitment, or directly through a pay-as-you-go online subscription.
Important—The price in R$ is merely a reference; this is an international transaction and the final price is subject to exchange rates and the inclusion of IOF taxes. An eNF will not be issued.
US government entities are eligible to purchase Azure Government services from a licensing solution provider with no upfront financial commitment, or directly through a pay-as-you-go online subscription.
Important—The price in R$ is merely a reference; this is an international transaction and the final price is subject to exchange rates and the inclusion of IOF taxes. An eNF will not be issued.
o3
o3 is a powerful reasoning model from the o-series of reasoning models, pushing the frontier across coding, math, science, and visual perception. It excels in complex queries requiring multi-faceted analysis and performs strongly in visual tasks like analyzing images, charts, and graphics. The model features a 200K token context window and has a knowledge cutoff of June 2024.
| Model | Pricing (1M Tokens) | Pricing with Batch API (1M Tokens) |
|---|---|---|
| o3 2025-04-16 |
Input: $- Cached Input: $- Output: $- |
N/A |
o4-mini
o4-mini is a compact, efficient, and cost-effective reasoning model from OpenAI's o-series. It excels in math, coding, and visual tasks. The model features a 200K token context window and has a knowledge cutoff of June 2024.
| Model | Pricing (1M Tokens) | Pricing with Batch API (1M Tokens) |
|---|---|---|
| o4-mini 2025-04-16 |
Input: $- Cached Input: $- Output: $- |
N/A |
GPT-4.1 series
GPT-4.1 series is a highly advanced general-purpose model with extensive world knowledge and an enhanced ability to understand user intent, making it particularly adept at creative tasks and agentic planning. The series features a 1 million token context window and has a knowledge cutoff of June 2024.
| Model | Pricing (1M Tokens) | Pricing with Batch API (1M Tokens) |
|---|---|---|
| GPT-4.1-2025-04-14 |
Input: $- Cached Input: $- Output: $- |
N/A |
| GPT-4.1-mini-2025-04-14 |
Input: $- Cached Input: $- Output: $- |
N/A |
| GPT-4.1-nano-2025-04-14 |
Input: $- Cached Input: $- Output: $- |
N/A |
GPT-4.5
GPT-4.5-preview is the latest general purpose model with deep world knowledge and better understanding of user intent that makes it good at creative tasks and agentic planning. The model has 128K context and an October 2023 knowledge cutoff.
| Model | Pricing (1M Tokens) | Pricing with Batch API (1M Tokens) |
|---|---|---|
| GPT-4.5-Preview-2025-02-27 Global |
Input: $- Cached Input: $- Output: $- |
N/A |
o1
o1 is the new reasoning model series for complex tasks. The model has 200K context and an October 2023 knowledge cut-off.
| Model | Pricing (1M Tokens) | Pricing with Batch API (1M Tokens) |
|---|---|---|
| o1 2024-12-17 Global |
Input: $- Cached Input: $- Output: $- |
N/A |
| o1 2024-12-17 US/EU – Data Zones |
Input: $- Cached Input: $- Output: $- |
N/A |
| o1 2024-12-17 Regional |
Input: $- Cached Input: $- Output: $- |
N/A |
| o1 preview 2024-09-12 Global |
Input: $- Cached Input: $- Output: $- |
N/A |
| o1 preview 2024-09-12 US/EU – Data Zones |
Input: $- Cached Input: $- Output: $- |
N/A |
| o1 preview 2024-09-12 Regional |
Input: $- Cached Input: $- Output: $- |
N/A |
Plan with the Pricing Calculator
o3 Mini
The o3 mini is the updated version of o1 mini model. o3-mini is a fast, cost-efficient reasoning model tailored to coding, maths, and science use cases.
The o3-mini model now boasts an expanded context input window of 200K tokens and a maximum output of 100K tokens, providing ample space for complex and detailed responses. The o1 mini model has 128K context input. Both o3 and o1 models have a knowledge cut-off of October 2023.
| Model | Pricing (1M Tokens) | Pricing with Batch API (1M Tokens) |
|---|---|---|
| o3 mini 2025-01-31 Global |
Input: $- Cached Input: $- Output: $- |
Input: $- Output: $- |
| o3 mini 2025-01-31-US/EU – Data Zones |
Input: $- Cached Input: $- Output: $- |
Input: $- Output: $- |
| o3 mini 2025-01-31 Regional |
Input: $- Cached Input: $- Output: $- |
N/A |
| o1-mini 2024-09-12 Global |
Input: $- Cached Input: $- Output: $- |
N/A |
| o1-mini 2024-09-12 US/EU – Data Zones |
Input: $- Cached Input: $- Output: $- |
N/A |
| o1-mini 2024-09-12 Regional |
Input: $- Cached Input: $- Output: $- |
N/A |
Plan with the Pricing Calculator
Computer-Using Agent (CUA)
The Computer-Using Agent (CUA) is a specialised AI model that allows AI to interact with graphical user interfaces (GUIs), navigate applications and automate multi-step tasks – all through natural language instructions. The CUA model can be used as a tool in the Responses API.
| Model | Pricing |
|---|---|
| computer-use-preview Global |
Input: $-/1M tokens Output: $-/1M tokens |
Built-in tools
The Responses API and the Assistants API enable seamless interaction with tools like computer use, code interpreter, function calling, and file search, making it easy for developers to build AI agents.
| Tool | Input |
|---|---|
| Computer Use (Responses API only) |
Input: $-/1M tokens Output: $-/1M tokens |
| File Search Tool Call (Responses API only) | $-/1K tool calls |
| File Search* | $-/GB of vector-storage per day (1 GB free) |
| Code Interpreter** | $-/session |
*GB refers to binary gigabytes, where 1 gb is 2^30 bytes.
**If your assistant calls Code Interpreter simultaneously in two different threads, this would create two Code Interpreter sessions (2 * $-). Each session is active by default for one hour, which means that you would only pay this fee once if your user keeps giving instructions to Code Interpreter in the same thread for up to one hour.
Inference cost (input and output) varies based on the GPT model used with each Assistant. If your assistant calls Code Interpreter simultaneously in two different threads, this would create two Code Interpreter sessions (2 * $-). Each session is active by default for one hour, which means that the price is for up to one hour of giving instructions to Code Interpreter in the same thread.
Realtime API
Featured in the Realtime API, the GPT-4o-Realtime-Preview supports multilingual speech-to-speech capabilities. Optimised for real-time, low-latency conversations, it enables natural interactions with minimal delay, ideal for chatbots and conversational AI. GPT-4o is the comprehensive, more powerful version designed for complex tasks, while GPT-4o Mini is a smaller, more affordable option ideal for simpler applications where cost-efficiency and speed are priorities.
| Model | Pricing (1M Tokens) |
|---|---|
| GPT-4o-Realtime-Preview-2024-12-17-Global |
Text Input: $- Cached Input: $- Output: $- Audio Input: $- Cached Input: $- Output: $- |
| GPT-4o-Realtime-Preview-2024-12-17-US/EU – Data Zones |
Text Input: $- Cached Input: $- Output: $- Audio Input: $- Cached Input: $- Output: $- |
| GPT-4o-Realtime-Preview-2024-12-17-Regional |
Text Input: $- Cached Input: $- Output: $- Audio Input: $- Cached Input: $- Output: $- |
| GPT-4o-Mini-Realtime-Preview-2024-12-17-Global |
Text Input: $- Cached Input: $- Output: $- Audio Input: $- Cached Input: $- Output: $- |
| GPT-4o-Mini-Realtime-Preview-2024-12-17-US/EU – Data Zones |
Text Input: $- Cached Input: $- Output: $- Audio Input: $- Cached Input: $- Output: $- |
| GPT-4o-Mini-Realtime-Preview-2024-12-17-Regional |
Text Input: $- Cached Input: $- Output: $- Audio Input: $- Cached Input: $- Output: $- |
| GPT-4o-Realtime-Preview-2024-10-01-Global |
Text Input: $- Cached Input: $- Output: $- Audio Input: $- Cached Input: $- Output: $- |
| GPT-4o-Realtime-Preview-2024-10-01-US/EU – Data Zones |
Text Input: $- Cached Input: $- Output: $- Audio Input: $- Cached Input: $- Output: $- |
| GPT-4o-Realtime-Preview-2024-10-01-Regional |
Text Input: $- Cached Input: $- Output: $- Audio Input: $- Cached Input: $- Output: $- |
Chat Completions API
Featured in the Chat Completions API, the GPT 4o-Audio-Preview model processes and generates audio content. It supports advanced features like speech recognition and audio synthesis, ideal for asynchronous speech interactions and sentiment analysis. GPT-4o is the comprehensive, more powerful version designed for complex tasks, while GPT-4o Mini is a smaller, more affordable option ideal for simpler applications where cost-efficiency and speed are priorities.
| Model | Pricing (1M Tokens) |
|---|---|
| GPT-4o-Audio-Preview-2024-12-17-Global |
Text Input: $- Output: $- Audio Input: $- Output: $- |
| GPT-4o-Audio-Preview-2024-12-17-US/EU – Data Zones |
Text Input: $- Output: $- Audio Input: $- Output: $- |
| GPT-4o-Audio-Preview-2024-12-17-Regional |
Text Input: $- Output: $- Audio Input: $- Output: $- |
| GPT-4o-Mini-Audio-Preview-2024-12-17-Global |
Text Input: $- Output: $- Audio Input: $- Output: $- |
| GPT-4o-Mini-Audio-Preview-2024-12-17-US/EU – Data Zones |
Text Input: $- Output: $- Audio Input: $- Output: $- |
| GPT-4o-Mini-Audio-Preview-2024-12-17-Regional |
Text Input: $- Output: $- Audio Input: $- Output: $- |
GPT-4o
GPT-4o is the most advanced multimodal model that’s faster and cheaper than GPT-4 Turbo with stronger vision capabilities. The model has 128K context and an October 2023 knowledge cutoff.
| Model | Pricing (1M Tokens) | Pricing with Batch API (1M Tokens) |
|---|---|---|
| GPT-4o-2024-1120 Global |
Input: $- Cached Input: $- Output: $- |
Input: $- Output: $- |
| GPT-4o-2024-1120 US/EU – Data Zones |
Input: $- Cached Input: $- Output: $- |
Input: $- Output: $- |
| GPT-4o-2024-1120 Regional |
Input: $- Cached Input: $- Output: $- |
N/A |
| GPT-4o-2024-08-06 Global |
Input: $- Cached Input: $- Output: $- |
Input: $- Output: $- |
| GPT-4o-2024-08-06 US/EU – Data Zones |
Input: $- Cached Input: $- Output: $- |
Input: $- Output: $- |
| GPT-4o-2024-08-06 Regional |
Input: $- Cached Input: $- Output: $- |
N/A |
| GPT-4o-2024-0513 Global |
Input: $- Output: $- |
Input: $- Output: $- |
| GPT-4o-2024-0513 US/EU – Data Zones |
Input: $- Output: $- |
N/A |
| GPT-4o-2024-0513 Regional |
Input: $- Output: $- |
N/A |
Plan with the Pricing Calculator
GPT-4o mini
GPT-4o mini is the most cost-efficient small model, and has vision capabilities. The model has 128K context and an October 2023 knowledge cutoff.
| Model | Pricing (1M Tokens) | Pricing with Batch API (1M Tokens) |
|---|---|---|
| GPT-4o-mini-0718 Global |
Input: $- Cached Input: $- Output: $- |
Input: $- Output: $- |
| GPT-4o-mini-0718 US/EU – Data Zones |
Input: $- Cached Input: $- Output: $- |
Input: $- Output: $- |
| GPT-4o-mini-0718 Regional |
Input: $- Cached Input: $- Output: $- |
N/A |
Plan with the Pricing Calculator
Provisioned
You can allocate and manage throughput for deployments, ensuring predictable performance and stable capacity. You are charged an hourly rate per model regardless of usage, but you can also secure additional savings through monthly and annual reservations. Discover how to transition your regional deployments and provisioned reservations to global and data zones on this Learn page.
| Model | Min PTUs | PTU Hourly pricing | PTU Monthly Reservation Pricing | PTU Yearly Reservation Pricing |
|---|---|---|---|---|
| GPT-4o Global | 15 | $- | $- | $- |
| GPT-4o US/EU Data Zones | 15 | $- | $- | $- |
| GPT-4o Regional | 50 | $- | $- | $- |
| Fine-Tuned GPT-4o-Regional | 50 | $- | $- | $- |
| GPT-4o Mini Global | 15 | $- | $- | $- |
| GPT-4o Mini US/EU Data Zones | 15 | $- | $- | $- |
| GPT-4o Mini Regional | 25 | $- | $- | $- |
| Fine-Tuned GPT-4o-Mini Regional | 25 | $- | $- | $- |
Plan with the Pricing Calculator
Base models
| Models | Usage per 1,000 tokens |
|---|---|
| Babbage-002 | $- |
| Davinci-002 | $- |
Fine-tuning models
| Model | Pricing | |
|---|---|---|
| GPT-4.1 | Regional |
Input: $-/1M tokens Cached Input: $-/1M tokens Output: $-/1M tokens Training: $-/1M tokens Hosting: $-/hour |
| Global |
Input: $-/1M tokens Cached Input: $-/1M tokens Output: $-/1M tokens Training: $-/1M tokens Hosting: $-/hour |
|
| GPT-4.1-mini | Regional |
Input: $-/1M tokens Cached Input: $-/1M tokens Output: $-/1M tokens Training: $-/1M tokens Hosting: $-/hour |
| Global |
Input: $-/1M tokens Cached Input: $-/1M tokens Output: $-/1M tokens Training: $-/1M tokens Hosting: $-/hour |
|
| GPT-4o-2024-08-06 | Regional |
Input: $-/1M tokens Cached Input: $-/1M tokens Output: $-/1M tokens Training: $-/1M tokens Hosting: $-/hour |
| Global |
Input: $-/1M tokens Cached Input: $-/1M tokens Output: $-/1M tokens Training: use regional Hosting: $-/hour |
|
| GPT-4o-mini | Regional |
Input: $-/1M tokens Cached Input: $-/1M tokens Output: $-/1M tokens Training: $-/1M tokens Hosting: $-/hour |
| Global |
Input: $-/1M tokens Cached Input: $-/1M tokens Output: $-/1M tokens Training: use regional Hosting: $-/hour |
|
| GPT-3.5-Turbo (16K) | Regional |
Input: $-/1M tokens Output: $-/1M tokens Training: $-/1M tokens Hosting: $-/hour |
| GPT-3.5-Turbo (4K) | Regional |
Input: $-/1M tokens Output: $-/1M tokens Training: $-/1M tokens Hosting: $-/hour |
Image models
| Models | Quality | Resolution | Price (per 100 images) |
|---|---|---|---|
| Dall-E-3 | Standard | 1024 * 1024 | $- |
| Standard | 1024 * 1792, 1792 * 1024 |
$- | |
| Dall-E-3 | HD | 1024 * 1024 | $- |
| HD | 1024 * 1792, 1792 * 1024 |
$- | |
| Dall-E-2 | Standard | 1024 * 1024 | $- |
Embedding models
| Models | Per 1,000 tokens |
|---|---|
| Ada | $- |
| text-embedding-3-large | $- |
| text-embedding-3-small | $- |
Speech Models
| Models | Price |
|---|---|
| Whisper | $-/hour |
| TTS (Text to Speech) | $-/1M characters |
| TTS HD | $-/1M characters |
Legacy Language Models
| Models | Context | Input (Per 1M Tokens) | Output (Per 1M Tokens) |
|---|---|---|---|
| GPT-3.5-Turbo-0301 | 4K | $- | $- |
| GPT-3.5-Turbo-0613 | 4K | $- | $- |
| GPT-3.5-Turbo-0613 | 16K | $- | $- |
| GPT-3.5-Turbo-1106 | 16K | $- | $- |
| GPT-3.5-Turbo-0125 | 16K | $- | $- |
| GPT-3.5-Turbo-Instruct | 4K | $- | $- |
| GPT-4-Turbo | 128K | $- | $- |
| GPT-4-Turbo-Vision | 128K | $- | $- |
| GPT-4 | 8K | $- | $- |
| GPT-4 | 32K | $- | $- |
Azure pricing and purchasing options
Connect with us directly
Get a walkthrough of Azure pricing. Understand pricing for your cloud solution, learn about cost optimisation and request a customised proposal.
Talk to a sales specialistSee ways to purchase
Purchase Azure services through the Azure website, a Microsoft representative or an Azure partner.
Explore your optionsAdditional resources
Azure OpenAI Service
Learn more about Azure OpenAI Service features and capabilities.
Pricing calculator
Estimate your expected monthly costs for using any combination of Azure products.
SLA
Review the Service Level Agreement for Azure OpenAI Service.
Documentation
Review technical tutorials, videos, and more Azure OpenAI Service resources.
Frequently asked questions
-
Azure OpenAI Service offers pricing based on both Pay-As-You-Go and Provisioned Throughput Units (PTUs). Pay-As-You-Go allows you to pay for the resources you consume, making it flexible for variable workloads. PTUs offers a predictable pricing model where you reserve and deploy a specific amount of model processing capacity. This model is ideal for workloads with consistent or predictable usage patterns, providing stability and cost control.
-
To learn more about PTUs and Azure OpenAI pricing please read PTU documentation or contact our sales specialist.
Talk to a sales specialist for a walk-through of Azure pricing. Understand pricing for your cloud solution.
Get free cloud services and a $200 credit to explore Azure for 30 days.