Models & Pricing

ASUME decides how to construct and maintain company understanding. Users do not select GPT, Claude, or another model for this process. Understanding is a maintained ASUME product state, and the systems used to build it are managed by ASUME.

Understanding usage is priced according to the information ASUME processes:

Build understanding applies when ASUME processes information to construct the initial understanding of a company. Maintain understanding applies when ASUME processes information later to keep that understanding current.

A page is one item ASUME reads and understands. Depending on the source, this may be a web page, a page within a PDF, or an eligible record from information a Customer has authorised ASUME to use. Different companies can therefore require different amounts of Understanding usage depending on the information available about them.

The models and systems ASUME chooses internally do not change the Customer's published Understanding rate. Choosing a different option in Ask ASUME also does not change how the underlying company understanding was built or maintained.

Ask ASUME works with the company understanding ASUME has already constructed and maintained. Model and answering choices happen at this layer.

For each question, you can use:

• Auto, which uses ASUME Router to choose the most appropriate model for the task.
• Corollary, ASUME's managed answering system for working with maintained company understanding.
• Or a supported third-party model, when you want direct control over the model used for the Ask.

At a simplified level:

Company information → Maintained understanding → Ask ASUME → Answer

Changing the option used for Ask ASUME changes how the answer is produced. It does not rebuild the company or create a separate underlying understanding.

Corollary is ASUME's managed answering system for Ask ASUME.

It works from maintained company understanding and uses supported third-party models behind the scenes. ASUME manages the underlying model selection, context preparation, and answering process, so you do not need to choose a provider or model yourself.

Corollary is designed for everyday Ask ASUME use where the user wants a strong balance of reasoning quality, speed, and cost without managing model selection directly.

Because ASUME controls how Corollary uses the underlying models, Corollary has its own published rate. The third-party model used internally for an individual Ask does not change the rate charged to the Customer.

All prices are per million tokens.

Corollary pricing reflects the managed ASUME answering system rather than the public API price of any single underlying model.

ASUME Router is the model-routing system behind Auto in Ask ASUME. It is not a model and not an answering system.

When you select Auto, the Router chooses an available model based on the question, the maintained company context involved, the reasoning required, expected speed, and cost.

A straightforward question can therefore be handled by a faster and more efficient model, while a more demanding question can be routed to a stronger model when the additional capability is useful.

ASUME Router does not have its own model rate. Auto usage is charged at the published ASUME rate of the model selected for that Ask.

The distinction is:

• Auto uses ASUME Router to choose a model for the Ask.
• Corollary is ASUME's managed answering system.
• GPT, Claude, and other supported models are third-party models the user can select directly.

Users who want more control can bypass Corollary and select a supported third-party model directly.

This can be useful when you prefer a particular provider, want a specific model's capabilities, need consistency around one model, want direct control over cost, or simply want to compare how different models answer the same company question.

Different models have different reasoning capabilities, context limits, speeds, and prices. A lower-cost model may be sufficient for a straightforward question, while a stronger frontier model may be appropriate for a more demanding task.

When a specific model is selected, Ask ASUME uses that model for the question and charges usage at the model's published ASUME rate. The choice affects the Ask only; it does not affect ASUME's maintained understanding of the company.

ASUME exposes a selected set of third-party models rather than every model available from every provider. The initial catalogue focuses on models that provide meaningfully different choices in capability, speed, and cost.

All prices are per million tokens.

The rates shown in ASUME documentation are the rates billed by ASUME for direct model usage. They may differ from the model provider's own API prices. Provider-specific pricing rules, such as long-context rates or cache-related charges, are described on the relevant model page where applicable.

For most users, Auto or Corollary should be the natural starting point.

Use Auto when you want ASUME Router to decide how the question should be answered. You ask the question, and the Router selects an available model.

Use Corollary when you specifically want ASUME's managed answering system and its predictable published pricing.

Choose a third-party model directly when the identity of the model matters to you or when you want explicit control over capability and cost.

These options give users control without requiring everyone to become familiar with the differences between every available model.

Each supported third-party model has its own documentation page. ASUME Router and Corollary also have dedicated pages explaining the system behind Auto and ASUME's managed answering system.

A third-party model page describes the model in the context of Ask ASUME rather than reproducing the provider's general-purpose documentation. It can include the provider, model identifier, context window, relative speed and cost, current ASUME pricing, strengths, limitations, and the kinds of Ask ASUME tasks for which the model may be particularly suitable.

For example, the GPT-5.6 Sol page can explain when its stronger reasoning may justify its higher cost, while a lower-cost model page can explain when that model is sufficient for simpler work.

The ASUME Router page explains the system behind Auto: how a model is selected for each Ask, and how usage is billed at that model's rate. Corollary's page explains ASUME's managed answering experience, how it works with maintained understanding, and why a user may prefer it to choosing an individual model.

Model availability and pricing may change as providers release, update, or retire models. The current ASUME documentation is the source of truth for supported options and their ASUME rates.

A company can be understood once and then queried many times through different answering options.

Suppose ASUME has already constructed and maintained an understanding of a company. One user may ask a question through Corollary, another may use GPT-5.6 Sol, and a later question may be answered using Claude Opus 5. All of those questions can operate on the same maintained company understanding.

The selected model or answering system affects how an individual Ask is answered. It does not create a separate company state.

Likewise, ASUME may process new information and update the maintained understanding even when nobody has asked a new question. That work belongs to Understanding usage and remains independent of Ask ASUME model usage.

This separation keeps responsibility clear: ASUME manages how the company is understood, while Ask ASUME provides different ways to use that understanding.

ASUME plans can include an allowance for Ask ASUME usage.

Because Corollary and third-party models have different prices, usage is value-based rather than treating every question as one identical unit. The same included allowance can therefore support more usage with a lower-cost option than with repeated use of a more expensive frontier model.

Two users can consequently ask the same number of questions while consuming different amounts of usage. A workspace that relies primarily on Corollary or lower-cost models may use its allowance differently from one that repeatedly selects premium models directly.

Understanding allowances remain separate. The amount of Understanding usage consumed depends on the information ASUME processes to build and maintain companies, not on which answering option is later used.

When included Ask ASUME usage is exhausted, a workspace can continue with on-demand usage where enabled or move to a plan with a larger allowance.

For Corollary or a directly selected model, on-demand usage is charged at the corresponding published rate. For Auto, the rate of the model selected by ASUME Router applies.

Team and Enterprise administrators can control whether additional usage is permitted and, where available, set workspace or organisation-level spending limits.

Reaching an included allowance does not require ASUME to silently replace a directly selected model with a cheaper one solely for billing purposes.

ASUME plans combine product access, Understanding capacity, Ask ASUME usage, collaboration, and administrative capabilities.

• Individual is designed for one person working with a focused set of companies and includes maintained understanding and Ask ASUME usage.
• Teams Standard adds shared company understanding, shared portfolios, team permissions, shared private evidence, centralised administration, usage visibility, and collaboration capabilities.
• Teams Premium is designed for teams with larger workloads or more advanced workflows. It provides higher included Understanding and Ask ASUME capacity together with capabilities such as API, MCP, advanced agents and automations, and faster refresh where available.
• Enterprise is designed for organisations requiring pooled usage, invoice or purchase-order billing, SSO or SAML, SCIM, audit logs, service accounts, advanced access controls, organisation-wide usage management, or dedicated support.

Current plan pricing and included allowances are listed on the ASUME Pricing page.

ASUME pricing has three distinct layers.

The plan determines the product capabilities, collaboration model, and included allowances available to the workspace.

Understanding usage covers the work required to construct and maintain company understanding. ASUME decides which models and systems are used for this process.

Ask ASUME usage covers how questions are answered against that understanding. Users can choose Corollary, select a supported third-party model directly, or use Auto so ASUME Router can choose the model.

The relationship is:

Company information → ASUME-managed Understanding

followed by:

Maintained understanding → Ask ASUME → Auto, Corollary, or selected model → Answer

This is the basis of ASUME's model and usage pricing. It keeps the underlying company understanding independent from the model or answering system used for any particular question.