> For the complete documentation index, see [/llms.txt](/llms.txt).

> Source: /docs/resources/faq

# FAQ

If you need help, feel free to reach out to us at the [Upstage help center](https://support.upstage.ai) or join our [Discord community](https://go.upstage.ai/discord). To check for any ongoing service incidents, visit [status.upstage.ai](https://status.upstage.ai/).

## Getting started

### Q. Can I test the API without signing up or providing payment information?

You can test the API for free—without logging in or registering a payment method—directly from the documentation.

- Use the [API Reference](/api/chat) or [Playground](https://console.upstage.ai/playground/chat) to interact with live endpoints.
- No login or credit card is required for basic testing.
- Example inputs and expected outputs are pre-filled to help you get started quickly.

### Q. Is the Playground free to use?

Yes, the Playground is free for all users. No login or payment method is required. However, any input may be used for model improvement, so avoid entering sensitive or personal data.

### Q. I'm having trouble logging in or signing up.

1. Please try signing up again after waiting for some time.
2. If the issue remains unresolved, please try signing up from a different location using a different device (either a PC or mobile).
3. If the issue is still not resolved, it may be due to your internet connection settings.
- We have found that the Ad Guard ad-blocker program can block the signup process. If you are using a similar program, please disable it and try the signup process again.

### Q. I closed the verification code screen before verifying my email.

You can finish verifying your email yourself — there is no need to sign up again.

1. Open the [Verify email](/verify-email) page. You can also find it under the menu next to the Log in button, or via the link at the bottom of the sign-in page.
2. Enter the email address you signed up with, then enter the verification code from the email you received. The code is valid for 24 hours after it is sent. If the code is missing or expired, click **Resend code** to get a new one.
3. Once verified, you can sign in to your account as usual.

## API Usage

### Authentication & Authorization

#### Q. Why am I getting '401 Unauthorized' error during an API request?

This issue occurs when authentication cannot be performed with the API key. Please check that you have correctly written the `Authorization` value in the header of the request.

```
curl -X POST https://api.upstage.ai/v1/document-digitization \ 
-H "Authorization: Bearer ${API_KEY}" \ # ex: xxxYYYzzzAAAbbbCCC
-F "image=@./image.png" \
-F "model=ocr"
```

⚠️ If the placeholder braces (`${}`) are left around your key, authentication may fail, leading to an error.

```
# Example of an error
curl -X POST https://api.upstage.ai/v1/document-digitization \
-H "Authorization: Bearer ${xxxYYYzzzAAAbbbCCC}" \
-F "image=@./image.png" \
-F "model=ocr"
```

⚠️ An authentication error may also occur if quotation marks `" "` are accidentally omitted or altered.

```
# Example of an error
curl -X POST https://api.upstage.ai/v1/document-digitization \
-H "Authorization: Bearer xxxYYYzzzAAAbbbCCC \
-F "image=@./image.png" \
-F "model=ocr"
```

⚠️ Please also check for typos, including trailing spaces.

```
# Example of an error
curl -X POST https://api.upstage.ai/v1/document-digitization \
-H "Authorization: Bearer x xxYYYzzzAAAbbbCCC " \
-F "image=@./image.png" \
-F "model=ocr"
```

⚠️ Do not delete the word `Bearer`.

```
# Example of an error
curl -X POST https://api.upstage.ai/v1/document-digitization \
-H "Authorization: xxxYYYzzzAAAbbbCCC" \
-F "image=@./image.png" \
-F "model=ocr"
```

### File Formats & Processing

#### Q. What file formats are supported for Document Parse?

Document Parse API supports commonly used image and document formats.

- **Supported file formats:** JPEG, PNG, BMP, PDF, TIFF, HEIC, DOCX, PPTX, XLSX, HWP, HWPX
- **Maximum file size:** 50MB

#### Q. What file formats are supported for OCR?

OCR API supports commonly used image and document formats.

- **Supported file formats:** JPEG, PNG, BMP, PDF, TIFF, HEIC, DOCX, PPTX, XLSX, HWP, HWPX
- **Maximum file size:** 50MB

#### Q. What's the simplest solution when HWP files fail to process?

Convert the file to PDF and re-upload it.

### Rate Limits & Performance

#### Q. Why am I getting 429 Too Many Requests error during an API request (rate limit)?

Rate limits are restrictions set by our API on how frequently a user can make requests within a given time frame. Rate limits vary by your commitment tier. For detailed information, please refer to the [rate limits page](/docs/guides/rate-limits). If you encounter 429 errors, reduce the frequency of your requests and try again after a few seconds or a minute. To raise the limit itself, see [How can I increase the rate limit, and are there any conditions required?](/docs/resources/faq#how-can-i-increase-the-rate-limit-and-are-there-any-conditions-required).

#### Q. How can I increase the rate limit, and are there any conditions required?

Rate limits are automatically determined by your commitment tier. To increase your rate limit, purchase credits to upgrade your commitment tier. You can choose a one-time purchase, or monthly or yearly auto-recharge, at the [credit purchase page](https://console.upstage.ai/billing/credit). For details on each tier and its rate limits, visit [upstage.ai/pricing/api](https://www.upstage.ai/pricing/api). If no commitment tier covers the limits you need, [contact us](https://www.upstage.ai/contact-us?utm_medium=console&utm_source=docs&utm_content=faq) with your requirements.

#### Q. Do rate limits apply per API key or per Organization?

Your Organization sets the limits, and each API key uses them on its own.

- The commitment tier of your Organization sets the rate limits. Every key in the Organization gets the same limits, including new keys.
- Requests are counted for each key separately. Requests from one key do not use up the limit of another key.
- For example, if your tier allows 100 RPM for a model and you have 3 keys, each key can send up to 100 requests per minute to that model.
- Each model has its own limit. Requests to one model do not count toward the limit of another model.
- The [rate limit response headers](/docs/guides/rate-limits#rate-limit-response-headers) show the remaining allowance of the key that sent the request.

This applies to model APIs such as Chat, Embeddings, Document Parse, Information Extract, and Document Classification.

#### Q. Why is the response time slow or delayed?

Temporary surges in usage can cause delays.

- Monthly availability is maintained at over 95% under our [Service Level Agreement](/docs/terms/sla)
- Check [status.upstage.ai](https://status.upstage.ai/) for real-time updates

#### Q. Can I request multiple documents at once?

We currently do not provide an interface for requesting multiple documents at once. We recommend sending images one at a time in series instead of sending multiple simultaneous requests. Sending many requests at once may result in a "429 Too Many Requests" error.

#### Q. How long does Async API processing take?

Processing time depends on queue status, document size, and complexity. During peak demand, requests may be delayed. For faster results, consider using the Sync API.

### Error Handling

#### Q. Why am I getting '403 Forbidden' error during an API request?

A 403 error may occur due to exhausted credits or failed payments.

**Common reasons:**

- No remaining credits
- Failed billing charge
- API calls are sent from an IP address that is blocked by internal firewall policies

**How to resolve:**

Please contact the [Upstage help center](https://support.upstage.ai) for assistance.

#### Q. What does '403 Client Error: Forbidden for url' mean in Document Parse API?

In the Document Parse Async API, inference results are stored for 30 days, but the download_url is only valid for about 15 minutes from the time it is issued. If you try to access a previously saved download_url after it has expired, a 403 error will occur.

**How to resolve:**

- Re-fetch the result using the original request_id
- The API will generate a new download_url in the response
- This re-fetch will not consume additional credits

If the issue persists after retrying, please contact the [Upstage help center](https://support.upstage.ai) for further assistance.

#### Q. What are the common error codes in Document Parse Async API?

Document Parse Async API can encounter errors in three scenarios:

**1. Request Errors (400, 413, 415):**

- **400 Bad Request**: Invalid model name or no document in request
- **413 Payload Too Large**: Document exceeds 50MB limit
- **415 Unsupported Media Type**: Unsupported document format

**2. Batch Inference Errors:**

- Occurs when processing fails during model inference
- Check the `batches` field in API response for `failed` status and `failure_message`

**3. Result Retrieval Failures:**

- Download URL generation issues (15-minute validity limit)
- Common with long documents requiring multiple batch URLs

#### Q. How do I handle batch inference errors in Document Parse Async API?

Batch inference errors occur when the engine fails to process document batches (10 pages each).

**How to identify:**

- Check the `batches` field in your API response
- Look for `"status": "failed"` entries
- Read the `failure_message` for error details

**Common causes:**

- Internal server errors during processing
- Document validation issues
- Resource constraints

**What to do:**

- Retry the request if it's a temporary issue
- Check document format and size requirements

#### Q. When does the 'Failed to process document' error occur in Async API?

- File name too long (recommend ≤300 characters for Korean, ≤900 for English).
- Document processing errors.
- Limitations due to specific document structures or formats.

#### Q. What should I do if a document remains stuck in 'started' status?

1. Shorten the file name (≤300 Korean characters, ≤900 English) and retry.
2. Note that some documents may require longer processing times, so waiting may help.

### Chat

#### Q. Why does a 500 error occur, and what should I do when it happens?

A 500 error is usually temporary.

- Monthly availability is maintained at over 95% under our [Service Level Agreement](/docs/terms/sla)
- Check [status.upstage.ai](https://status.upstage.ai/) for service incidents

#### Q. Why am I getting 429 Too Many Requests error during an API request (rate limit)?

Rate limits are restrictions set by our API on how frequently a user can make requests within a given time frame. Rate limits vary by your commitment tier. For detailed information, please refer to the [rate limits page](/docs/guides/rate-limits). If you encounter 429 errors, reduce the frequency of your requests and try again after a few seconds or a minute. To raise the limit itself, see [How can I increase the rate limit, and are there any conditions required?](/docs/resources/faq#how-can-i-increase-the-rate-limit-and-are-there-any-conditions-required).

#### Q. How can I increase the rate limit, and are there any conditions required?

Rate limits are automatically determined by your commitment tier. To increase your rate limit, purchase credits to upgrade your commitment tier. You can choose a one-time purchase, or monthly or yearly auto-recharge, at the [credit purchase page](https://console.upstage.ai/billing/credit). For details on each tier and its rate limits, visit [upstage.ai/pricing/api](https://www.upstage.ai/pricing/api). If no commitment tier covers the limits you need, [contact us](https://www.upstage.ai/contact-us?utm_medium=console&utm_source=docs&utm_content=faq) with your requirements.

#### Q. Why is the response time slow or delayed?

Temporary surges in usage can cause delays.

- Monthly availability is maintained at over 95% under our [Service Level Agreement](/docs/terms/sla)
- Check [status.upstage.ai](https://status.upstage.ai/) for real-time updates

#### Q. What should I do when the Chat API response format isn't what I expected?

Specify the desired output format with an example in your prompt.

- Chat API supports flexible formats (e.g., JSON, Markdown, list). Refer to the [API Reference page](/api/chat) for more details.
- Including an output example in the prompt improves consistency

#### Q. How can I calculate the number of tokens used in LLM?

Token usage in LLM can be calculated using the guide in the documentation.

- Refer to: Docs → Guides → [Counting tokens](/docs/guides/counting-tokens)
- Token cost is based on model type and token length

### Document parsing

#### Q. What is the difference between OCR and Document Parse?

Document Parse understands the structure and layout of documents to extract text, tables, and images, while OCR mainly focuses on text recognition.

#### Q. What does '403 Client Error: Forbidden for url' mean in Document Parse API?

In the Document Parse Async API, inference results are stored for 30 days, but the download_url is only valid for about 15 minutes from the time it is issued. If you try to access a previously saved download_url after it has expired, a 403 error will occur.

**How to resolve:**

- Re-fetch the result using the original request_id
- The API will generate a new download_url in the response
- This re-fetch will not consume additional credits

If the issue persists after retrying, please contact the [Upstage help center](https://support.upstage.ai) for further assistance.

#### Q. What are the common error codes in Document Parse Async API?

Document Parse Async API can encounter errors in three scenarios:

**1. Request Errors (400, 413, 415):**

- **400 Bad Request**: Invalid model name or no document in request
- **413 Payload Too Large**: Document exceeds 50MB limit
- **415 Unsupported Media Type**: Unsupported document format

**2. Batch Inference Errors:**

- Occurs when processing fails during model inference
- Check the `batches` field in API response for `failed` status and `failure_message`

**3. Result Retrieval Failures:**

- Download URL generation issues (15-minute validity limit)
- Common with long documents requiring multiple batch URLs

#### Q. How do I handle batch inference errors in Document Parse Async API?

Batch inference errors occur when the engine fails to process document batches (10 pages each).

**How to identify:**

- Check the `batches` field in your API response
- Look for `"status": "failed"` entries
- Read the `failure_message` for error details

**Common causes:**

- Internal server errors during processing
- Document validation issues
- Resource constraints

**What to do:**

- Retry the request if it's a temporary issue
- Check document format and size requirements

#### Q. What file formats are supported for Document Parse?

Document Parse API supports commonly used image and document formats.

- **Supported file formats:** JPEG, PNG, BMP, PDF, TIFF, HEIC, DOCX, PPTX, XLSX, HWP, HWPX
- **Maximum file size:** 50MB

#### Q. Where are base64-encoded images in Document Parse output?

They are not displayed directly in HTML but included in the `elements` array of the response. See the [API Reference page](/api/parse/document-parsing) for details.

#### Q. How do I fix table recognition issues when parsing receipts with Document Parse?

Complex documents such as borderless receipts can be difficult to recognize. Document Parse provides `standard`, `enhanced`, and `auto` modes, and `enhanced` is intended for documents with complex tables, images, and charts. Some cases still require post-processing.

See [Document parsing](/docs/capabilities/parse/document-parsing) for how to set the mode. Per-page rates differ by mode; see [upstage.ai/pricing/api](https://www.upstage.ai/pricing/api).

#### Q. What's the simplest solution when HWP files fail to process?

Convert the file to PDF and re-upload it.

#### Q. How long does Async API processing take?

Processing time depends on queue status, document size, and complexity. During peak demand, requests may be delayed. For faster results, consider using the Sync API.

#### Q. When does the 'Failed to process document' error occur in Async API?

- File name too long (recommend ≤300 characters for Korean, ≤900 for English).
- Document processing errors.
- Limitations due to specific document structures or formats.

#### Q. What should I do if a document remains stuck in 'started' status?

1. Shorten the file name (≤300 Korean characters, ≤900 English) and retry.
2. Note that some documents may require longer processing times, so waiting may help.

### Document OCR

#### Q. What is the difference between OCR and Document Parse?

Document Parse understands the structure and layout of documents to extract text, tables, and images, while OCR mainly focuses on text recognition.

#### Q. Why am I getting 429 Too Many Requests error during an API request (rate limit)?

Rate limits are restrictions set by our API on how frequently a user can make requests within a given time frame. Rate limits vary by your commitment tier. For detailed information, please refer to the [rate limits page](/docs/guides/rate-limits). If you encounter 429 errors, reduce the frequency of your requests and try again after a few seconds or a minute. To raise the limit itself, see [How can I increase the rate limit, and are there any conditions required?](/docs/resources/faq#how-can-i-increase-the-rate-limit-and-are-there-any-conditions-required).

#### Q. How can I increase the rate limit, and are there any conditions required?

Rate limits are automatically determined by your commitment tier. To increase your rate limit, purchase credits to upgrade your commitment tier. You can choose a one-time purchase, or monthly or yearly auto-recharge, at the [credit purchase page](https://console.upstage.ai/billing/credit). For details on each tier and its rate limits, visit [upstage.ai/pricing/api](https://www.upstage.ai/pricing/api). If no commitment tier covers the limits you need, [contact us](https://www.upstage.ai/contact-us?utm_medium=console&utm_source=docs&utm_content=faq) with your requirements.

#### Q. Why is the response time slow or delayed?

Temporary surges in usage can cause delays.

- Monthly availability is maintained at over 95% under our [Service Level Agreement](/docs/terms/sla)
- Check [status.upstage.ai](https://status.upstage.ai/) for real-time updates

#### Q. Can I request multiple documents at once?

We currently do not provide an interface for requesting multiple documents at once. We recommend sending images one at a time in series instead of sending multiple simultaneous requests. Sending many requests at once may result in a "429 Too Many Requests" error.

#### Q. What file formats are supported for OCR?

OCR API supports commonly used image and document formats.

- **Supported file formats:** JPEG, PNG, BMP, PDF, TIFF, HEIC, DOCX, PPTX, XLSX, HWP, HWPX
- **Maximum file size:** 50MB

#### Q. Does the OCR API provide line-level output?

The OCR API does not return a dedicated `line` field. Line-level output shown in Playground is a post-processing result. You can reconstruct it using bounding box coordinates in the `words` array and line breaks in `pages.text`.

#### Q. How can I fix line breaks and spacing errors in OCR results?

The OCR API does not currently offer options to adjust text correction. Recognition errors in dense or complex layouts are known limitations and will be gradually improved with future model updates.

#### Q. How can I improve OCR text recognition quality?

1. Change OCR options (`ocr=force` vs. `ocr=auto`).
2. Test the latest model version.
3. For complex layouts, apply structured prompts or post-processing.

### Universal extraction

#### Q. Is there an interface for filtering specific fields in the Information Extraction API response?

We do not provide an interface to filter specific fields from the API response. We recommend that you extract the necessary field from the response JSON.

```
# Example for filtering "store.store_address" from the receipt extraction API reponse
response = requests.post(url, headers=headers, files=files)
data = response.json()
addresses = [field['value'] for field in data.get('fields', []) if (field['key'] == 'store.store_address' and field['type'] == 'content')]
print(addresses)
```

#### Q. What is the role of keys in Universal Information Extract?

Keys are not just labels. Defining clear and specific keys improves extraction accuracy. For example: "Patient Name", "Treatment Date".

## Organization & Members

### Q. Do rate limits apply per API key or per Organization?

Your Organization sets the limits, and each API key uses them on its own.

- The commitment tier of your Organization sets the rate limits. Every key in the Organization gets the same limits, including new keys.
- Requests are counted for each key separately. Requests from one key do not use up the limit of another key.
- For example, if your tier allows 100 RPM for a model and you have 3 keys, each key can send up to 100 requests per minute to that model.
- Each model has its own limit. Requests to one model do not count toward the limit of another model.
- The [rate limit response headers](/docs/guides/rate-limits#rate-limit-response-headers) show the remaining allowance of the key that sent the request.

This applies to model APIs such as Chat, Embeddings, Document Parse, Information Extract, and Document Classification.

### Q. What is an Organization?

An Organization is the unit that owns your API keys, billing, usage, and limits. Every Upstage Console user automatically belongs to a default Organization and is its Owner. Members you invite share that Organization's API access and billing.

### Q. I already have an account and API keys. Do I need to change anything?

No. Your existing API keys, payment method, credits, and monthly postpaid billing limit move to your default Organization automatically, and you remain the Owner. Your keys keep working without any change.

### Q. What is the difference between an Owner and a Member?

- **Owner:** Manages the Organization — inviting and removing Members, changing settings, and managing billing and usage limits.
- **Member:** Creates and manages their own API keys, and can view the Organization’s billing, usage, and limit information without being able to change it.

An Organization has exactly one Owner.

### Q. How do I invite Members to my Organization?

Only the Organization Owner can invite Members. No subscription is required.

1. Go to `Members` in your Organization.
2. Enter one or more email addresses, separated by line breaks, commas, or spaces.

Everyone who accepts joins as a Member.

If the Console behaves differently from this, please contact the [Upstage help center](https://support.upstage.ai).

### Q. I invited someone, but they didn't receive anything.

Invitations are handled inside the Console, and no email is sent. Please let the person know directly.

- To accept, they sign in to Upstage Console, open the Organization switcher at the top of the sidebar, and select `Invitations`.
- We do not check whether the address already has an account, so an invitation is created even for an address that has not signed up yet. In that case the invitation appears only after they sign up with the same email address.

### Q. Someone invited me to their Organization. Where do I accept?

Open the Organization switcher at the top of the Console sidebar and select `Invitations`. From there you can accept or decline.

- The menu appears with a count badge only while you have a pending invitation.
- You are not a Member and have no access to the Organization until you accept.

### Q. Do invitations expire?

Yes. An invitation expires 7 days after it is sent and disappears from the list. There is no resend action — the Owner can simply invite the person again.

### Q. How are API keys tied to an Organization?

Each API key belongs to the Organization selected when you created it, and works only there.

- Usage and cost are billed to that Organization. A key created in a different Organization is billed there instead.
- Model access, pricing, rate limits, and the monthly postpaid billing limit are set per Organization. All keys follow the same settings, including new ones.
- Each key uses the rate limits on its own. For example, if you use separate keys for development and production, heavy development traffic does not use up the limit of the production key. For details, see [Do rate limits apply per API key or per Organization?](/docs/resources/faq#do-rate-limits-apply-per-api-key-or-per-organization).
- If you belong to more than one Organization, switching Organizations changes the key list you see.
- Owners see every key in the Organization, while Members see only the keys they created.
- You cannot delete an Organization at this time. To stop using a key in another Organization, revoke it on the **API keys** page there.

### Q. How many API keys can I create?

Every Member, including Owners, can have up to 10 API keys in each Organization. Revoked keys do not count toward the limit.

- The Organization has no total limit. For example, an Organization with 5 Members can have up to 50 keys.
- You cannot restore a revoked key. If you reach the limit, revoke a key that you no longer use on the **API keys** page, then create a new key.

### Q. Can I see usage for each API key?

Yes. On the Usage page, filter usage by API key or by the key owner's email.

1. Click the account menu (your profile icon) at the top right of the Console
2. Click **Manage organization**
3. If you belong to more than one organization, select the organization at the top of the sidebar
4. In the sidebar, click **Usage**
5. In the **Usage Details** section, select a key in **API key**. To see one person's keys, enter their email in **Owner email**
6. Set **Start (UTC)** and **End (UTC)**, then click **Apply**. One search covers up to 92 days
7. To save the result, click **Download CSV**. The file uses the same filters and dates

### Q. What happens to API keys when a Member leaves the Organization?

API keys that the Member created in that Organization are revoked immediately, whether the Owner removes them or they leave on their own. Applications using those keys will fail to authenticate.

- Usage and billing history is preserved.
- A revoked key cannot be reactivated — issue a new one from the API keys menu in the Organization.
- Keys in other Organizations are not affected.

### Q. What can a Member see?

Members have read-only access to the Organization's payment method summary, credit balance, invoices, Commitment tier, total spend, and monthly postpaid billing limit. Changing any of them is Owner-only.

For per-key usage and cost, Members see only the keys they created, while Owners see all keys in the Organization.

### Q. Can I create more than one Organization?

Not at this time. Every user has one default Organization, and you can additionally join other Organizations by accepting an invitation.

## Billing & Usage

### Pricing & Plans

#### Q. Which models are available for free?

As part of the [AWS Initiative](https://www.upstage.ai/events/ai-initiative-2026-en), Document Parse (DP) and Solar Pro (LLM) models are free for up to one year for nonprofit, educational, and research institutions. The Information Extract (IE) model has transitioned to a paid plan.

#### Q. What is the pricing for the Upstage API?

Upstage API pricing can be found on the official Pricing page.

- The Upstage API uses a pay-as-you-go or prepaid credit model
- Pricing varies by model (e.g., Solar Pro, Document Parse)

[See pricing details](https://www.upstage.ai/pricing/api)

#### Q. How do I use Studio vouchers for a hackathon?

1. Log in to Studio with the email address used to request the vouchers.
2. As the organization Owner, open Settings → Billing and check your subscription vouchers.
3. Choose the plan covered by the vouchers. For a team, set the billing seats to the total number of members, including the Owner, and apply enough vouchers for all seats at once.
4. Confirm that the amount due is zero before confirming the change. No payment method is needed when vouchers cover the full amount.

If vouchers are missing, check the email address and organization, then contact the person who issued your vouchers. If the amount due is not zero, do not confirm the change; ask the person who issued your vouchers to check it first.

#### Q. Will my hackathon Studio plan charge my card after it ends?

A Studio plan activated entirely with event vouchers does not automatically renew by charging your card. When the voucher-funded subscription ends and no usable vouchers remain, it returns to the Free plan.

This does not apply to a separately started paid subscription, a card-paid trial conversion, or a plan partly paid by card. If you used one of those options, check the renewal settings in Studio → Settings → Billing instead of assuming the event benefit cancels card billing.

#### Q. Can I get a custom enterprise plan?

Yes! Enterprise plans include custom rate limit, dedicated support, SLA guarantees, and additional features. Contact our [sales team](https://www.upstage.ai/contact-us) to discuss your specific requirements.

### Billing & Charges

#### Q. Are failed API requests charged?

Requests that do not return a valid response are generally not billed. If a system error caused an incorrect charge, please provide the request ID or timestamp so we can issue a refund.

#### Q. How are the charges for the service billed?

Billing is based on either postpaid (pay-as-you-go) or prepaid (credits).

- **Pay-as-you-go:** Billed monthly by usage. This means that charges are only incurred for the paid services actually used. The cost for the paid services used is calculated monthly from the first to the last day of each month and is billed through the payment method registered on a specified date.
- **Credits:** Preloaded, usage deducted

#### Q. Does a Studio voucher also cover Console API usage?

No. Studio subscription vouchers and Console API billing are separate. A Studio voucher does not make API calls free.

Before calling APIs, check whether benefits such as AWS AI Initiative or API credits apply, including eligible models, limits, and expiration dates. Without applicable benefits or credits, register a payment method to use paid APIs. Registering a card alone does not create API usage charges; paid API usage does.

#### Q. What is the refund policy?

Purchased credits and Studio subscriptions can be refunded within 7 days of payment if they have not been used.

- **Refundable:** Within 7 days of payment, with no usage
- **Non-refundable:** Credits or subscriptions that have already been used, and credits provided free of charge

* You must agree to the refund policy before purchase. Refer to [Articles 16 and 20 of the Terms of Service](https://www.upstage.ai/terms-of-service).

* To request a refund, log in to the Console first so we can verify that you own the account. Then contact us in one of these ways:

- The chat button at the bottom right of the screen
- **Feedback** at the bottom of the side menu

### Payment Methods

#### Q. Is there an immediate charge after adding a payment method?

No, you will not be charged immediately.

1. You will not be billed upon registering a payment method as long as you have free credits available.
2. Charges are incurred solely based on your API usage, so you won't face any fees when not using it.

#### Q. How can I change my payment method?

1. Go to the [Payment page](https://console.upstage.ai/billing/payment).
2. In the **Payment method** section, click **Change**.

From there, you can register a new payment method.

#### Q. I want to remove the registered payment method.

Payment method removal is handled by support, not a self-service delete button.

1. Log in to [Upstage Console](https://console.upstage.ai) with the account you need help with.
2. Accept cookies so the chat widget appears, then open the chat at the bottom right.
3. Ask a support agent to remove the registered payment method while you are logged in.

Removing a payment method and changing a Studio subscription’s renewal settings are separate requests. Check your renewal settings as well if you have a paid subscription.

#### Q. What payment methods are supported?

We currently support credit card payments.

- Visa, Mastercard, and other major cards accepted
- Card information is securely encrypted
- Payment methods can be updated anytime
- **Other methods:** If you need a payment option other than credit card, please contact the [Upstage help center](https://support.upstage.ai)

### Credits & Budget

#### Q. How can I set a monthly postpaid billing limit?

The monthly postpaid billing limit caps how much you can keep using pay-as-you-go on your payment method each month after your credits run out. It applies to charges remaining after credits are deducted, not to usage covered by credits. Only the Org Owner can set or change it.

1. Click the account menu (your profile icon) at the top right of the Console
2. Click **Manage organization**
3. If you belong to more than one organization, select the organization at the top of the sidebar
4. In the sidebar, click **Payment**
5. In the **Monthly postpaid billing limit** section, click **Set**. If a limit is already set, click **Change**
6. Choose what happens when credits run out: **Stop API requests** (use credits only), **Continue up to a monthly limit** (enter the amount), or **Continue without a cap**
7. Click **Save**
- To remove the limit, click **Remove** or choose **Continue without a cap**, then confirm
- The limit applies per calendar month and resets on the 1st at 00:00 UTC. Usage within the limit is billed on the 1st of the following month
- API requests are blocked only when you have no credits left and this month’s postpaid charges have reached the limit. While credits remain, reaching the limit doesn’t block requests
- You can set the limit before adding a payment method, and it applies once you add one. Without a payment method, API requests stop when credits run out
- Enforcement may take up to 1 second, so minor overages can still be billed

#### Q. What are credits?

Credits are a prepaid balance system for Upstage AI API services. Even if your credits run out, your API usage won't be interrupted—charges will be billed to your registered payment method. By pre-purchasing credits, you can also receive various bonus benefits.

Credits apply only to usage after you purchase them. See [Do credits apply to usage before I purchased them?](/docs/resources/faq#do-credits-apply-to-usage-before-i-purchased-them)

#### Q. Do credits apply to usage before I purchased them?

No. Credits apply only to usage after you purchase them and within their valid period. Usage before the purchase is billed to your registered payment method on the 1st of the next month.

For example, if you purchase credits on November 19:

- Usage from November 1 to 18 is billed to your payment method on December 1.
- Usage from November 19 is deducted from your credits.

You can see the same example on the [Billing guide](https://console.upstage.ai/billing) page in the Console.

#### Q. Is there a minimum purchase amount?

No minimum purchase is required. You can recharge any amount you want.

*** Commitment Benefits:** Monthly purchases of $100 or more unlock bonus credits and higher rate limits. For details, visit the [Pricing page](https://www.upstage.ai/pricing/api).

#### Q. How do I purchase credits?

1. Go to [Billing → Credit](https://console.upstage.ai/billing/credit) in the Console
2. Click Purchase credits
3. Enter your desired amount or choose a preset (+$100, +$1,000, +$5,000, +$10,000)
4. Check the box to confirm how credits apply and the refund policy
5. Click "Confirm and pay" to complete your purchase

### Auto-recharge & Commitment

#### Q. What is the Commitment tier?

It's a tiered system that provides benefits based on your purchase amount:

- **Bonus Credits:** Additional credits from +10% to +20% depending on purchase size (+30% to +40% on yearly plans)
- **Rate Limit Benefits:** Higher requests per second and more pages processed (Tier 1–3)
- **Automatic Tier Management:** Explore, Build, Scale—adjusted automatically by monthly purchase
- **Eligibility:** Applies when monthly purchases are $100 or more
- **Learn more:** [Pricing page](https://www.upstage.ai/pricing/api)

#### Q. What is the validity period of Commitment bonus credits?

Bonus credits from a one-time purchase or monthly auto-recharge can be used for 1 month from the date they are granted. Bonus credits from yearly auto-recharge can be used for 1 year. Unused bonus credits expire after that.

You can check the expiry date of each credit in the **Valid period (UTC)** column under History on the [Billing → Credit](https://console.upstage.ai/billing/credit) page.

#### Q. How do I set up auto-recharge?

You can enable auto-recharge from the [Billing → Credit](https://console.upstage.ai/billing/credit) page by clicking Start auto-recharge.

- **Amount:** Set your preferred amount (bonus included)
- **Payment type:** Choose Monthly auto-recharge or Yearly auto-recharge
- **Credit recharge date:** Pick your preferred date
- **Preview (Commitment Benefits):** Preview bonuses and tier benefits before confirming

#### Q. How do I terminate auto-recharge?

On the [Billing → Credit](https://console.upstage.ai/billing/credit) page, click **Turn off** in the **Auto-recharge** section, then click **Turn off** again to confirm. Only the next scheduled recharge will be cancelled; your existing credits will remain in your account.

#### Q. Do I receive bonus credits with auto-recharge?

Yes. The larger the recharge amount, the bigger the bonus you receive. For detailed information, visit the [Pricing page](https://www.upstage.ai/pricing/api).

### Issues & Troubleshooting

#### Q. Why does my Studio invitation say “The Studio subscription has no available seat”?

The organization does not have enough subscription seats. Receiving vouchers alone does not add seats: the Owner must apply the plan with enough billing seats for the whole team, including the Owner.

If the Owner already started the plan with fewer seats than the team needs, adding seats takes more vouchers than the number of seats added. The current plan is credited only for its remaining value, so adding 1 seat usually takes 2 vouchers. If the amount due is not zero, do not confirm the change; contact the person who issued your vouchers instead.

After the Owner updates the seats, accept the same invitation again using the account it was sent to. Team members do not need to activate their own vouchers to join the team subscription.

#### Q. I would like to request a refund due to unexpected charges incurred.

1. Visit the [Upstage help center](https://support.upstage.ai).
2. Choose the `Developer support` → `Console billing` under support topics.
3. Fill in the specific details you need assistance with.
4. Click `Send`.

## Security & Privacy

### Q. How secure is your API and console?

Security is our top priority. Upstage is certified for **SOC 2**, **HIPAA**, and **ISO/IEC 27001 & 27701**, ensuring our systems meet rigorous international standards for information security and privacy. All API traffic is encrypted over HTTPS, requires API key authentication, and is subject to rate limiting. We also perform regular internal and third-party security audits.

### Q. What is your uptime guarantee?

We guarantee **monthly availability of over 95%** for all production API services (excluding the free tier), as stated in our [Service Level Agreement](/docs/terms/sla). If availability falls below this level due to our fault, eligible customers may receive credit compensation.

### Q. Do you store my documents or API inputs?

We handle data storage differently depending on the service:

- **Real-time API**: Input/output data is processed in memory and not stored.
- **Asynchronous API**: Input documents are stored until inference completes (up to 3 days). Output data is stored for up to 30 days to support feature delivery. Data is **never used** for service R&D or AI model training.
- **Playground/Demo Services**: Uploaded documents and input data may be stored and used for service improvement, AI technology R&D, and legal compliance, and may be retained until you deregister your account.
- **Third-party AI services**: When using APIs provided via a third-party AI service, your data is processed in accordance with that provider's policy.

### Q. Is my data used to train your models?

- **API (real-time & asynchronous)**: No. We do **not** use API input/output data for service R&D or AI model training.
- **Console (Playground & demo services)**: Inputs/uploads **may be used** for service improvement and AI technology R&D (per Article 22) and may be retained until you deregister your account.
- **Third-party AI services**: When using APIs provided via a third-party AI service, your data is processed in accordance with that provider's policy.

### Q. Where is my data processed?

Upstage processes data in secure environments in accordance with internal policies and applicable laws. If you have specific data residency requirements, please contact the [Upstage help center](https://support.upstage.ai).

### Q. Are you GDPR compliant?

Upstage complies with global privacy standards, including **ISO/IEC 27701** (Privacy Information Management) and relevant local laws such as the Korean Personal Information Protection Act (PIPA). We follow GDPR principles and provide data processing agreements (DPAs), support data export, and honor deletion requests. For jurisdiction-specific coverage, please review our [Privacy Policy](https://www.upstage.ai/privacy-policy).

### Q. Can I delete uploaded documents?

Yes. You may request document deletion at any time through our data access/deletion procedure, and files will be permanently erased immediately.

### Q. How can I delete my data?

If you wish to delete stored documents, parsed data, project records, or your entire account data, please contact the [Upstage help center](https://support.upstage.ai). We will verify your request and process it promptly.

### Q. I would like to deactivate my account and discontinue service usage.

You can delete your account directly from the [Account page](https://console.upstage.ai/account).

## Platform & Infrastructure

### Q. Can I deploy Upstage models on my own infrastructure?

Yes! All models can be deployed on-premises, and most are available through cloud marketplaces.

**Deployment options:**

- **Upstage Console:** Direct API access for all models
- **Cloud Marketplace:** AWS Marketplace (most models) and Azure Marketplace (Information Extract only)
- **On-premises:** Gives you full control within your own infrastructure

**For detailed guides:** Visit [deployment options page](/docs/deployment-options), [AWS guide](/docs/deployment-options/aws/document-intelligence-ai-agent), or [Azure guide](/docs/deployment-options/azure/managed-applications). Contact our [sales team](https://www.upstage.ai/contact-us) for enterprise deployment discussions.

### Q. What are the requirements for deploying Information Extract on Azure Marketplace?

Azure Marketplace deployment requires specific GPU quota and configuration.

**Key requirements:**

- **GPU quota:** 32 vCPUs for Standard NVADSA10v5 Family (request increase before deployment)
- **Regions:** 9 US regions + Southeast Asia, Korea Central
- **Network:** Choose private VNet or public IP setup
- **JIT access:** Recommended for Upstage support troubleshooting

**Important:** VM size is pre-configured, managed resource group should not be altered. For detailed instructions, see [Azure Information Extract guide](/docs/deployment-options/azure/managed-applications) or contact the [Upstage help center](https://support.upstage.ai).

## Technical Support

### Q. How can I receive technical support from Upstage?

Technical support from Upstage can be requested through the [Upstage help center](https://support.upstage.ai).