Get started for free

Model Catalog

What this section is for

The model catalog allows managing available AI models in the system - adding custom models, configuring their parameters and pricing, controlling availability for users.

How to access this section

  • Through main menu: "Models"
  • Direct link: /models

Main features

Model list

  • View all added models with pagination
  • Search and filtering by parameters
  • Status management (active/inactive models)
  • Editing parameters of existing models
  • Deletion of unnecessary models

Adding new models

  • Creating custom model configurations
  • Pricing setup for different token types
  • Provider selection from supported platforms
  • Visibility control (private, corporate, public)

Catalog interface

Page header

  • Navigation: Home β†’ Models
  • "Add Model" button - for creating new model
  • Loading indicator during operations

Model list

Each model is displayed as a card with following information:

Main information

  • Model name - technical name (e.g., gpt-4-turbo)
  • Visibility badge - access level (Private, Corporate, Public)
  • Provider badge - AI platform (OpenAI, Anthropic, AWS, etc.)

Pricing information

  • Input Price - cost of input tokens
    • Calculated price with markup
    • Base price in brackets
  • Output Price - cost of output tokens
    • Calculated price with markup
    • Base price in brackets

Model actions

  • Activate/Deactivate - toggle availability status
  • Edit button ✏️ - modify parameters
  • Delete button πŸ—‘οΈ - delete model (with confirmation)

Pagination

  • Items per page: configurable (default 10)
  • Page navigation with numbers and buttons
  • Auto-reset to first page when changing size

Creating and editing models

Add model modal

Opens when clicking "Add Model" button with new model creation form.

Edit model modal

Opens when clicking on model or edit button.

Form fields

Main parameters

  • Provider (required)

    • Provider dropdown list
    • Options: OpenAI, Anthropic, Mistral, AWS, Flutch and their combinations
    • Determines which platform will be used

  • Model Name (required)

    • Technical model name
    • Examples: gpt-4-turbo, claude-3-sonnet, mistral-large
    • Must exactly match provider API

Pricing

  • Base Input Price (required)

    • Cost per 1 million input tokens
    • Numeric field with decimal support
    • Step: 0.001

  • Base Output Price (required)

    • Cost per 1 million output tokens
    • Usually more expensive than input tokens
    • Numeric field with step 0.001

Default AI parameters

  • Default Temperature

    • Slider from 0 to 2 with step 0.1
    • Default: 0.7
    • Controls response creativity

  • Default Max Tokens

    • Slider from 1 to 4096
    • Default: 2048
    • Maximum response length

Availability settings

  • Active (toggle)

    • Default: enabled
    • Inactive models unavailable for use
    • Can be deactivated without deletion

  • Visibility (required)

    • Private: only for creator
    • Corporate: for entire organization
    • Public: for all system users

Form management

  • "Cancel" - cancel without saving
  • "Create Model" / "Update Model" - save changes
  • Loading indication during request processing

Available providers

Main providers

  • OpenAI - GPT models (GPT-4, GPT-3.5)
  • Anthropic - Claude models (Claude-3, Claude-2)
  • Mistral - Mistral AI models
  • AWS - models through AWS Bedrock

Flutch providers

  • Flutch - platform's own models
  • Flutch OpenAI - OpenAI through Flutch
  • Flutch Anthropic - Anthropic through Flutch
  • Flutch Mistral - Mistral through Flutch

Visibility levels

Private

  • Available only to model creator
  • Used for testing and personal projects
  • Not visible to other organization users

Corporate

  • Available to all organization users
  • Standard level for working models
  • Allows controlling company expenses

Public

  • Available to all system users
  • Rarely used, mainly for demonstration
  • Requires careful pricing consideration

Interface states

Catalog loading

During loading, card skeletons are displayed (4 pieces or by screen size).

Empty catalog

  • Message: "No models found. Create your first model to get started."
  • "Add Model" button to create first model

Model operations

  • Loading indicators on action buttons
  • Interface blocking during operations
  • Delete confirmation through system dialog

Main usage scenarios

Adding new model

  1. Click "Add Model"
  2. Select provider (e.g., OpenAI)
  3. Enter model name (e.g., gpt-4-turbo)
  4. Configure pricing (base prices from provider)
  5. Set default AI parameters
  6. Choose visibility level (Corporate for working models)
  7. Click "Create Model"

Pricing configuration

  1. Find needed model in list
  2. Click on it or press edit button
  3. Change Base Input Price and Base Output Price
  4. If needed, adjust Temperature and Max Tokens
  5. Save changes

Managing model availability

  1. Find model that needs deactivation
  2. Click "Deactivate" button
  3. Model becomes unavailable for creating new agents
  4. Existing agents with this model continue working
  5. If needed, reactivate with "Activate" button

Deleting outdated models

  1. Ensure model isn't used in active agents
  2. Click delete button πŸ—‘οΈ
  3. Confirm deletion in dialog
  4. Model will be permanently deleted

Integration with agents

Model selection when creating agent

  • Only active models with appropriate visibility level available
  • Models displayed with provider and cost indication
  • Default settings automatically applied

Cost calculation

  • System uses set prices to calculate agent costs
  • Real cost may differ from base due to markups
  • Display both base and calculated costs

Limitations

Technical limitations

  • Cannot change provider of existing model
  • Cannot delete model used in active agents
  • Maximum Temperature (2.0) and Max Tokens (4096) values

Business limitations

  • Deactivated models unavailable for new agents
  • Private models visible only to creator
  • Price changes don't affect already spent funds

Interface

  • One provider at a time when creating model
  • No bulk model editing
  • No ability to copy settings between models

Common problems and solutions

Model doesn't appear when creating agent

Causes: inactive, inappropriate visibility level, incorrect name Solution: check model status, visibility level, API name accuracy

Error when saving model

Causes: incorrect name, duplication, provider issues Solution: check name uniqueness, parameter correctness

Incorrect usage cost

Cause: pricing errors, uncounted provider markups Solution: recalculate base prices, account for provider fees

Cannot delete model

Cause: model used in active agents Solution: find and update agents to different model, then delete

Security and monitoring

Access control

  • Only administrators can manage catalog
  • Visibility levels limit usage
  • Audit of all model changes

Usage monitoring

  • Track model popularity
  • Control costs by models
  • Analyze pricing effectiveness

Error protection

  • Validation of all form parameters
  • Confirmation of destructive operations
  • Option to deactivate instead of delete

Related sections