Pricing policy

1. Introduction

AI Hub is available in three tiers: a community offering available for general use, a commercial offering suitable for companies and organizations, and an enterprise offering for organizations with advanced needs. For plan comparisons and pricing, see Instabase pricing.

This Unit Pricing Policy (“Policy”) sets forth the terms and conditions under which you may purchase and use consumption units (“Consumption Units” or “Units”) in AI Hub. By purchasing or using Consumption Units, you agree to be bound by this Policy and the AI Hub Terms of Use unless otherwise specified in an Instabase Order Form.

2. Purchase and value of Consumption Units

You may purchase 100 Consumption Units per U.S. dollar, in increments of 100 Units.

3. Usage of Consumption Units

You may choose from available model options and utilize Consumption Units for specific services and apps available on AI Hub as follows:

Converse

Converse automatically uses the best model for each query:

• Queries to single documents use the Advanced Model.

• Queries to multiple documents use the Multistep Model.

When making queries to single or multiple documents, you can enable Research mode. Research mode supports complex reasoning prompts and uses a more powerful but sometimes slower variant of the Multistep model. Research mode follows the Multistep model’s pricing.

For example, if making a query in Converse to a single 10-Page document:

• Converse uses the Advanced Model and you are charged 5 Units/query.

• If you enable Research mode, Converse uses the Multistep Model and you are charged 10 Units/query.

For queries to multiple documents, the pricing is based on the total number of pages for all documents in the message’s scope. For example, for a multiple document query in Converse to three 30-Page documents (90 Pages in total):

• Converse uses the Multistep Model and you are charged 50 Units/query.

• If you enable Research mode, Converse uses the Multistep Model and you are charged 50 Units/query.

Chatbots

A chatbot query represents a query to all files in the chatbot’s knowledge base, defined as the documents present in the chatbot’s origin conversation. Query pricing is based on the total number of pages of all documents in the knowledge base and the model used.

Chatbots automatically use the best model for each query:

• Queries to chatbots with a knowledge base consisting of a single document use the Advanced Model.

• Queries to chatbots with a knowledge base consisting of multiple documents use the Multistep Model.

When making chatbot queries, you can enable Research mode. Research mode supports complex reasoning prompts and uses a more powerful but sometimes slower variant of the Multistep model. Research mode follows the Multistep model’s pricing.

For example, if making a query to a chatbot with a knowledge base of a single 100-Page document:

• The chatbot uses the Advanced Model and you are charged 30 Units/query.

• If you enable Research mode, the chatbot uses the Multistep Model and you are charged 50 Units/query.

For example, if making a query to a chatbot with a knowledge base of three 10-Page documents (30 Pages in total):

• The chatbot uses the Multistep Model and you are charged 36 Units/query.

• If you enable Research mode, the chatbot uses the Multistep Model and you are charged 36 Units/query.

Build and Build app runs

AI Hub uses a field rate to calculate Build and Build app run costs. The length of the document and the model used determine the field rate. The total cost depends on the number of fields in your schema.

• Each field extracted from a document counts as one field in the schema. You can use field-level model selection to choose which model is used to extract a given field. Field rate varies between models.

• If your Build project includes classification or split classification, classification is counted as one additional field in your schema. Classification is charged per document classified, at the Standard model field rate.

  You are not charged for classification when a document is classified as Other, instead of as one of your pre-defined classes.

For example, if you choose the Standard Model to extract all fields in Build or when running a Build app:

• For a 2-Page pay stub, you are charged 1 Unit/field for extracting fields such as net pay, pay period, employee name, etc.

  • With classification enabled, you are charged an additional 1 Unit. Classification of a document counts as one additional field, charged at the Standard model field rate.

For example, if you choose the Advanced Model to extract all fields in Build or when running a Build app:

• For an 8-Page letter of credit, you are charged 10 Units/field for extracting fields such as description of goods, port of loading, etc.

  • With classification enabled, you are charged an additional 5 Units. Classification of a document counts as one additional field, charged at the Standard model field rate.

For example, if you use field-level model selection in Build or when running a Build app:

• For a 15-Page document, you are charged 7 Units/field for fields extracted with the Standard Model. You are charged 14 Units/field for fields extracted with the Advanced Model.

  • With classification enabled, you are charged an additional 7 Units. Classification of a document counts as one additional field, charged at the Standard model field rate.

When using split classification with a multi-document packet, the field rate is based on the length of each document identified in the packet, not the total length of the document packet. This means when using split classification, you can expect the same pricing as if you sent in each document in the packet independently.

• For example, for a 200-Page document packet comprising four 50-Page bank statements, the field rate is based on the 50-Page length of the individual documents in the packet. For each 50-Page document, you are charged:

  • 15 Units for classification.

  • 15 Units/field for fields extracted with the Standard Model.

  • 30 Units/field for fields extracted with the Advanced Model.

Special apps

Some apps on AI Hub might also specify custom pricing per page processed or per document processed. If an app has a custom pricing model, it is specified clearly on the app page.

4. Page definition

• For documents that support pages (PDF, Word, PPT, TIFF images, etc.), each page in the document counts as 1 Page.

• For images (such as .png, .jpeg, and .bmp files), each image counts as 1 Page; multiple images embedded on a single page count as 1 Page.

• For email (.eml) files, where there is no native concept of a “page”, Instabase uses “Logical Page Count”. When calculating Logical Page Count for email files, 1 Page is defined as up to 10,000 characters, including white spaces. For example, if an email contains 8,500 characters of extracted text and white spaces, the email counts as 1 Page.

  • If an email contains attachments, those attachment files are automatically added to the project and they are billed according to their file type.

  • If an email contains inline images, and the images contain text, that text is extracted as part of the email. Text extracted from inline images contributes to character count when calculating Logical Page Count.

  • If an email is multi-threaded, meaning there is a chain of replies in the email, all characters across all threads in the email are included when calculating Logical Page Count.

• For spreadsheet files, including .csv, .xls, and .xlsx files, where there is no native concept of a “page”, Instabase uses Logical Page Count. When calculating Logical Page Count for spreadsheet files, 1 Page is up to 1,000 cells, including white spaces.

  • When defining the number of cells in one tab of a spreadsheet, Instabase considers the area defined by the uppermost left-hand cell containing data and the lowermost right-hand cell containing data. The total number of cells within that area, including any white spaces within the area, is included when calculating Logical Page Count.

  • If a spreadsheet file contains multiple tabs, each tab of the spreadsheet that contains data is included when calculating Logical Page Count.

5. Consumption Unit details

Consumption Units provide a convenient way to track the amount of AI Hub Services you have purchased and consumed. All purchases of Consumption Units are final and non-refundable. Consumption Units are non-transferable and can be used only by the customer that purchased them for the use of AI Hub apps, including Converse, Build, and other prebuilt apps that will be released over time. The Units are of no direct monetary value and cannot be redeemed for cash or cash equivalents. Subject to your compliance with the Policy, you subscribe to the amount of AI Hub usage you purchase as Consumption Units for up to twelve (12) months from the date of purchase (“Subscription Period”). Subsequent purchases of additional Consumption Units do not roll over, change or otherwise extend the Subscription Period; each purchase maintains its own Subscription Period. You can track your Consumption Units and their corresponding Subscription Periods in your AI Hub account. Units are consumed from oldest to newest.

6. International customers

International customers may purchase and use Consumption Units in accordance with this Policy. Additional local laws and regulations, however, might apply to customers located outside the United States. International customers are responsible for complying with all applicable local laws and regulations, including any tax obligations, currency conversion rules, and consumer protection requirements.

7. Taxes

Unless otherwise specified, the Consumption Units pricing does not include federal, state, local, and foreign taxes, duties, and other similar assessments (“Taxes”). You are responsible for all Taxes associated with your purchase, excluding Taxes based on our net income. You agree to timely pay such Taxes as applicable.

8. Notice of changes to the policy

Instabase reserves the right to update or modify this Policy from time to time. Any changes to the Policy will be communicated to customers through the Instabase AI Hub platform and effective upon publication except for any price increase, which will be effective thirty (30) days after notice.

9. Governing law

This Policy will be governed by and interpreted in accordance with the laws of the State of California, without giving effect to any principles of conflict of laws. Any legal action or proceeding arising under or related to this Policy will be brought exclusively in the federal and state courts located in San Francisco, California and the parties irrevocably consent to personal jurisdiction and venue in such courts.

10. Contact information

If you have any questions or concerns about this Policy, please contact Instabase at: ar@instabase.com