Azure OCR costs about $1.50 per 1,000 pages for plain text, and it costs that on both of the services people mean when they say "Azure OCR". Structure is the expensive part: Layout and every prebuilt model are $10.00 per 1,000 pages, and custom extraction is $30.00. Every figure here is pulled from Microsoft's published pricing pages and the public Azure Retail Prices API, with the meter name and effective date shown.
Written for US teams costing out an OCR vendor. Rates are East US pay-as-you-go and Microsoft varies them by region, so confirm on the current Azure pricing page. Last updated August 2026.
Upload a document to extract
Drop files here or click to upload
Up to 50 files
Free plan extracts the first 5, rest can be unlocked after
Uploading...
Before you pick a meter, drop in the document you meant to test on Azure and see what finished, validated fields look like.
Reading text off a page on Azure costs about $1.50 per 1,000, and it lands on that number whichever route you take. Azure AI Document Intelligence bills prebuilt-read at $1.50 per 1,000 pages, dropping to $0.60 above a million pages a month. Azure AI Vision bills its Read visual feature at $1.50 per 1,000 transactions, with the same break at a million. The price only moves when you ask for structure. Tables, selection marks and paragraph roles mean Layout, and Layout is billed as a prebuilt model at $10.00 per 1,000 pages, the same as invoice extraction. Train your own schema and you are at $30.00. So the real Azure OCR question is never "how much is OCR", it is "which meter does the thing I actually need sit on", and Microsoft answers that with model names rather than prices.
There is no product called Azure OCR. There are two services that read text, and a retired API that still shows up in old tutorials and old code. Microsoft's OCR overview names the split directly: the current Read editions are "OCR for images (version 4.0)" in Azure AI Vision and the "Document Intelligence read model", and the page carries a warning covering the legacy OCR API v3.2 and RecognizeText v2.1, adding that "There are no further updates to Azure Vision v3.2."
| Azure AI Vision | Azure AI Document Intelligence | |
|---|---|---|
| What you are reading | A photo, sign, poster or product label | A scanned or digital document |
| Microsoft's service | Azure AI Vision (Image Analysis 4.0) | Azure AI Document Intelligence |
| The identifier you call | Read visual feature | prebuilt-read |
| API style | Synchronous | Asynchronous, poll the operation |
| Billing unit | Transaction | Page |
| Standard rate | $1.50 per 1,000 | $1.50 per 1,000 |
| Free allowance | 5,000 transactions a month (F0) | 500 pages a month (F0) |
| Returns tables and key-value pairs | No | Yes, on prebuilt-layout and the prebuilt models |
| Microsoft's own steer | General, non-document images | "Use Document Intelligence Read OCR" |
The engine is shared, which is why the rate is the same. Microsoft describes Document Intelligence as including "a document-optimized version of Read as its OCR engine while delegating to other models for higher-end insights". What differs is the packaging: Vision hands you text and boxes through a synchronous call, Document Intelligence hands you a document object through an operation you poll. If you have already read our note on the difference between prebuilt-read and prebuilt-layout, this is the layer above it.
Azure AI Vision does not price OCR as a line item. It prices Image Analysis features in groups, and the two text features sit in different groups at different prices. The feature literally named OCR is in Group 1. The feature named Read is in Group 2, alongside Describe, Caption and Dense Captions. Group 1 is a third cheaper, and Group 1 is the one Microsoft's own documentation tells you not to build on.
| Image Analysis group | 0 to 1M | 1M to 10M | 10M to 100M | 100M+ |
|---|---|---|---|---|
| Group 1, including the feature named OCR | $1.00 | $0.65 | $0.60 | $0.40 |
| Group 2, including the feature named Read | $1.50 | $0.60 | $0.60 | $0.60 |
Per 1,000 transactions, East US, pay-as-you-go. Meters "Image Analysis Group 1 Transactions" and "Image Analysis Group 2 Transactions" on the Azure Retail Prices API, effective 2021-01-01, checked 17 August 2026. Group 2 has a single break at one million, so the last two columns repeat its 1M+ rate. The free F0 instance covers 5,000 transactions a month at 20 transactions per minute.
Saving $0.50 per 1,000 pages by calling the Group 1 OCR feature means building on the v3.2 API that Microsoft has frozen. The overview page states that all future Read OCR enhancements go into the two current services and that there are no further updates to v3.2. On a million pages a year that choice is worth $500, against the cost of migrating off a dead endpoint later. Take the $1.50.
This is the table most people are looking for when they search Azure OCR pricing. The meter name column matters more than it looks: it is what shows up on your Azure invoice, and it is the only reliable way to work out after the fact which model a workload actually billed on.
| What you are doing | Model identifier | Meter on your invoice | Per 1,000 pages | Volume break | Effective |
|---|---|---|---|---|---|
| Read | prebuilt-read |
S0 Read Pages | $1.50 | $0.60 above 1M pages a month | 2022-03-01 |
| Layout, and every prebuilt model | prebuilt-layout, prebuilt-invoice, prebuilt-receipt, prebuilt-idDocument, prebuilt-tax.us.w2 |
S0 Pre-built Pages | $10.00 | No published volume break | 2021-08-01 |
| Document classification | Custom classifier |
S0 pages for doc classifier | $3.00 | No published volume break | 2023-09-01 |
| Custom extraction | Your trained model |
S0 Custom Pages | $30.00 | No published volume break | 2024-06-01 |
| Custom generative extraction | Your generative model |
S0 Custom Generative Pages | $30.00 | No published volume break | 2024-06-01 |
| Add-ons (High Resolution, Font, Formula) | Request feature flags |
S0 Add-on for Pages | $6.00 | Charged on top of the model | 2024-11-01 |
| Query fields | queryFields |
S0 pages for query fields | $10.00 | Charged on top of the model | 2024-11-01 |
| Custom model training | Training run |
S0 Training | $3.00 per hour | Billed by the hour, not the page | 2024-11-01 |
East US pay-as-you-go, from the public Azure Retail Prices API, checked 17 August 2026. The free F0 instance covers 500 pages a month and, in Microsoft's words, "Does not support premium features, and does not include Query Meter."
Microsoft's pricing page has one line reading "All Prebuilt Models: Document, Layout, Receipt, Invoice, ID, W-2, 1098 Tax forms, Health insurance card, Contract" with a single price against it. Layout is not a flag on Read, it is a prebuilt model, so the moment you need a table you move from the $1.50 meter to the $10.00 meter. That is a 6.7 times jump, and it is the single biggest cost decision in an Azure document pipeline. The same shape shows up on AWS, where asking Textract for tables moves you off the $1.50 text meter too.
Microsoft's own billing FAQ says "When a document is submitted for analysis all pages are analyzed unless a page range is specified with the pages parameter in the request." Teams routinely post a 40-page loan file to pull two fields off page 3 and pay for 40 pages. Passing a page range costs one line of code and cuts that call by 95 percent. Add-ons compound the same way: High Resolution, Font and Formula each add $6.00 per 1,000 pages on top of whatever model you called, and they are easy to leave switched on after a test.
No. Azure prices batch identically to synchronous, meter for meter, to the cent. That is worth stating plainly because it runs against the instinct every engineer brings from LLM APIs, where batch is routinely half price. On Azure, batch changes how you submit work, not what it costs.
| Model | Synchronous meter | Per 1,000 | Batch meter | Per 1,000 | Difference |
|---|---|---|---|---|---|
| Read | S0 Read Pages | $1.50 | S0 Batch Read Pages | $1.50 | Identical |
| Prebuilt models | S0 Pre-built Pages | $10.00 | S0 Batch Pre-built Pages | $10.00 | Identical |
| Layout | Billed as prebuilt | $10.00 | S0 Batch Layout Pages | $10.00 | Identical |
| Document classification | S0 pages for doc classifier | $3.00 | S0 Batch Document Classifier Pages | $3.00 | Identical |
| Custom extraction | S0 Custom Pages | $30.00 | S0 Batch Custom Extraction Pages | $30.00 | Identical |
| Custom generative | S0 Custom Generative Pages | $30.00 | S0 Batch Custom Generative Pages | $30.00 | Identical |
| Add-ons | S0 Add-on for Pages | $6.00 | S0 Batch Add-On Pages | $6.00 | Identical |
| Query fields | S0 pages for query fields | $10.00 | S0 Batch Pages for Query Fields | $10.00 | Identical |
Every pair read off the Azure Retail Prices API on 17 August 2026. "S0 Batch Layout Pages" is the newest of these meters, with an effective date of 1 January 2026. There is more on how this compares with other vendors in our note on whether batch OCR costs less.
Only above a break-even, and the break-even is very different by model. A commitment tier is a fixed monthly fee, charged whether you use the pages or not, so the number that matters is how much of the tier you have to consume before it beats metered pricing. Read tiers pay off at 40 to 69 percent utilization. Prebuilt tiers need 75 to 95 percent, which is a much harder promise to make about next month's volume.
| Model | Tier | Monthly fee | Effective per 1,000 | Break-even vs pay-as-you-go | Utilization needed |
|---|---|---|---|---|---|
| Read | 500,000 pages | $375 | $0.75 | 250,000 pages | 50% |
| Read | 2,000,000 pages | $1,200 | $0.60 | 800,000 pages | 40% |
| Read | 8,000,000 pages | $4,200 | $0.53 | 5,500,000 pages | 69% |
| Read | 16,000,000 pages | $7,200 | $0.45 | 10,500,000 pages | 66% |
| Prebuilt | 20,000 pages | $190 | $9.50 | 19,000 pages | 95% |
| Prebuilt | 100,000 pages | $900 | $9.00 | 90,000 pages | 90% |
| Prebuilt | 500,000 pages | $4,000 | $8.00 | 400,000 pages | 80% |
| Prebuilt | 1,000,000 pages | $7,500 | $7.50 | 750,000 pages | 75% |
| Custom extraction | 20,000 pages | $540 | $27.00 | 18,000 pages | 90% |
| Custom extraction | 100,000 pages | $2,400 | $24.00 | 80,000 pages | 80% |
| Custom extraction | 500,000 pages | $10,500 | $21.00 | 350,000 pages | 70% |
| Custom extraction | 1,000,000 pages | $18,000 | $18.00 | 600,000 pages | 60% |
Monthly fees from the Azure Retail Prices API, East US, checked 17 August 2026. Break-even is the page count at which metered pricing costs the same as the fixed fee, using $1.50 per 1,000 for Read up to a million pages and $0.60 above it, $10.00 for prebuilt and $30.00 for custom extraction. Read break-evens rise at the top end because metered Read has already fallen to $0.60 by then.
Comparing the connected-container commitment tiers with the Azure cloud tiers on the same feed gives three different discounts, and they are exact on every tier. Read runs 10 percent cheaper in a container. Custom extraction runs 15 percent cheaper. Prebuilt models run 20 percent cheaper. So the case for running Azure in your own infrastructure is strongest precisely where the pages are most expensive, which is not the way most teams assume it works. Our note on on-premise OCR software covers what those containers can and cannot do.
Disconnected container tiers are quoted per year rather than per month, which makes them look like a different product. They are not. The Read 2,000,000-page disconnected tier is $12,960 a year, which is exactly twelve times the $1,080 connected monthly fee. Prebuilt at 100,000 pages is $8,640 a year against $720 a month. Custom at 100,000 is $24,480 against $2,040. The rate is identical on every tier we checked. What you are buying with disconnected is the air gap, not a discount.
Metered pricing, no commitment tier, before support and egress. The Read column includes the drop to $0.60 above a million pages, which is why five million pages costs less than five times what one million costs.
| Volume | Read (text only) | Layout or prebuilt | Custom extraction |
|---|---|---|---|
| 10,000 pages a month | $15 | $100 | $300 |
| 50,000 pages a month | $75 | $500 | $1,500 |
| 250,000 pages a month | $375 | $2,500 | $7,500 |
| 1,000,000 pages a month | $1,500 | $10,000 | $30,000 |
| 5,000,000 pages a month | $3,900 | $50,000 | $150,000 |
Rounded, East US, August 2026. What these numbers leave out is the part that usually costs more than the API: the queue, the retry logic, the confidence threshold, the screen a human uses to fix the 4 percent that came back wrong, and the export into whatever system needs the data. For a cross-vendor view of the same question, see OCR pricing per 1,000 pages and the buyer's OCR API pricing comparison.
The full rate card for the document service, with worked examples.
What happened to the old name, and which meter a Form Recognizer resource bills on.
The third Azure service in this space, priced per 1,000 pages.
File size, page counts and rate limits, separately from cost.
Which operation on which cloud returns what, and on which meter.
AWS, Azure, Google and the rest, normalized per 1,000 pages.
A meter table cannot tell you whether the fields come out right. Upload one of your real documents, look at what comes back, and then decide whether you are buying an OCR call or a finished pipeline.
