Introduction to credits / consumption

Written By Stanislas

Last updated 5 days ago

Learn how Swiftask credits work, what consumes them, and how different operations affect your credit balance.

Overview

Swiftask uses a credit-based system to measure your AI usage. Every AI interaction—from chat messages to document processing—consumes credits based on the complexity of the task and the resources required.

Understanding how credits work helps you make informed decisions about which AI models to use, how to structure your workflows, and how to optimize your spending. This guide explains what credits are, what consumes them, and how different factors affect your credit usage.

Prerequisites

To understand your credit consumption, you should:

  • Have an active Swiftask account

  • Be familiar with basic Swiftask features (Chat, Agents, Knowledge base)

  • Have access to your account settings to view credit usage

No technical knowledge is required—this guide is designed for all Swiftask users.

What are credits?

Credits are Swiftask's universal currency for AI operations. Think of credits as units that measure how much AI processing power you use.

Every AI operation consumes credits:

  • Sending a chat message

  • Generating an image or document

  • Processing uploaded files

  • Indexing documents in your knowledge base

  • Using agent skills and tools

  • Querying databases or APIs through agents

Your monthly credit allocation depends on your subscription plan. Once you understand what consumes credits, you can manage your usage more effectively.

How credits are calculated

Credit consumption follows a simple formula:

Base cost: A fixed cost per operation (e.g., starting a chat, creating an agent)

Token cost: Varies based on:

  • The AI model you're using

  • The length of your input message

  • The length of the AI's response

Feature cost: Additional costs for:

  • File attachments and document processing

  • Knowledge base operations

  • Agent skills and tools

  • Integrations and connectors

AI model pricing

Different AI models have different credit costs. More powerful models cost more per interaction.

Available models and their costs

Key takeaway: Swiftask default agent is the most cost-effective for simple tasks. Claude 3 Opus is the most expensive but offers the highest quality for complex work.

Chat message credit consumption

Basic chat message

When you send a message in Chat, credits are consumed based on:

  • Your input message length

  • The AI's response length

  • The AI model selected

With conversation history

AI includes previous messages for context. Longer conversations consume more credits because the AI processes the entire history with each new message.

Tip: Longer conversations cost more. Start a new chat when the conversation becomes too long or changes topic.

Viewing credit consumption in chat

After each AI response, you can see how many credits were consumed.

As shown in the image above, credit consumption appears at the bottom of each AI response (e.g., "259 credits"). This helps you understand the cost of each interaction in real-time.

File processing and document analysis

When you attach files to a chat message or upload documents to your knowledge base, Swiftask processes them and consumes credits based on file size and analysis complexity.

File attachment costs

What consumes credits during file processing:

  • Text extraction from PDFs

  • OCR (optical character recognition) for scanned documents and images

  • Table and structure recognition

  • Data parsing from spreadsheets (CSV, XLSX)

  • Code analysis from programming files

Knowledge base and document indexing

Adding content to your knowledge base consumes credits. The cost depends on:

  • The size of the documents you upload

  • The type of content (text files, PDFs, web pages, etc.)

  • The complexity of processing required

What consumes credits:

  • Uploading documents to your knowledge base

  • Importing web pages and websites

  • Syncing files from cloud storage (Google Drive, SharePoint, Dropbox)

  • Re-indexing updated documents

Tip: Upload documents to your knowledge base once and reuse them across multiple agents instead of attaching the same files repeatedly to chat messages.

Agent operations

When you use AI agents, credit consumption includes:

  • The AI model cost (same as chat)

  • Agent operation costs

  • Any skills or tools the agent uses

Agent costs

Agents consume credits each time they respond to a query. The cost depends on:

  • The AI model powering the agent

  • Whether the agent searches a knowledge base

  • Which skills or tools the agent uses (database queries, API calls, integrations)

Typical agent interaction:

  • Simple agent query: Lower cost

  • Agent with knowledge base: Moderate cost

  • Agent with multiple skills: Higher cost

Agent skills and tools

When agents use skills (integrations, database queries, API calls), each skill operation consumes additional credits.

Common skill operations:

  • Database queries

  • REST API calls

  • External integrations (CRM, project management tools, etc.)

  • Web searches

  • File generation

Tip: Only enable skills your agent actually needs. Each active skill can increase credit consumption.

Factors that affect credit consumption

1. AI model selection

The AI model you choose has the biggest impact on credit costs. Premium models like Claude Sonnet 4.5 cost significantly more than Swiftask default agent.

2. Message length

Longer input messages and longer AI responses consume more tokens and therefore more credits.

Recommendation: Be clear and concise in your prompts. Avoid unnecessary context or repetition.

3. Conversation history

AI includes previous messages for context. The longer the conversation, the more credits each new message consumes.

Recommendation: Start a new chat when conversations become too long or change topics significantly.

4. File attachments

Attaching files—especially large PDFs or images—increases credit consumption significantly.

5. Knowledge base size

Agents connected to large knowledge bases may consume more credits as they search through more content to find relevant information.

Recommendation: Keep knowledge bases focused. Use multiple smaller data sources instead of one massive collection.

6. Agent skills and tools

Each skill or tool your agent uses consumes additional credits.

Recommendation: Only enable skills that are necessary for your agent's specific tasks.

Practical use cases

Optimize for routine tasks

Use Swiftask default agent for everyday tasks like:

  • Quick questions and answers

  • Simple text generation

  • Basic document summaries

  • Routine agent interactions

Reserve premium models (GPT-4, Claude) for:

  • Complex analysis

  • Creative writing

  • Detailed research

  • Critical decision-making

Manage file processing costs

Before uploading large files:

  • Check the file size (smaller files = fewer credits)

  • Consider splitting large documents into smaller chunks

For knowledge base:

  • Upload documents once and reuse them across multiple agents

  • Update existing documents instead of creating duplicates

Control conversation length

Start fresh when:

  • The conversation exceeds 20-30 messages

  • You're changing topics significantly

  • The AI starts referencing old, irrelevant context

Keep conversations focused:

  • Ask one question at a time

  • Avoid repeating information the AI already has

  • Use clear, direct language

Tips & best practices

Choose the right model for the task

Don't use premium models for simple tasks. Match the model to the complexity of your work.

Keep messages concise

Longer messages consume more tokens. Be clear and direct in your prompts to reduce unnecessary token usage.

Use knowledge base instead of repeated attachments

If you frequently reference the same documents, add them to your knowledge base once instead of attaching them to every chat message.

Monitor your usage regularly

Check your credit usage dashboard weekly to identify patterns and optimize your workflows.

Start new conversations when needed

Long conversation histories increase credit consumption with each new message. Start fresh when conversations become too long.

Disable unnecessary agent skills

Only enable skills your agent actually needs. Each active skill can consume additional credits.

Troubleshooting

Issue: Higher than expected credit usage

Possible causes:

  • Using premium AI models for simple tasks

  • Long conversation histories

  • Large file attachments in ADVANCED mode

  • Agents with many active skills

Solutions:

  1. Check your credit usage dashboard to identify which operations consumed the most credits

  2. Switch to Swiftask default agent for routine tasks

  3. Start new conversations instead of continuing very long ones

  4. Review and disable unnecessary agent skills

Issue: Unexpected credit consumption for a single message

Possible causes:

  • Large file attachment

  • Very long AI response

  • Agent using multiple skills or database queries

Solutions:

  1. Check the credit indicator at the bottom of the AI response

  2. Review the file size if you attached documents

  3. Check if the agent used any skills or tools (visible in the response)

Issue: Credits depleted mid-month

Possible causes:

  • Underestimated monthly usage

  • Increased activity or new use cases

  • Inefficient workflows consuming more credits than necessary

Solutions:

  1. Review your credit usage dashboard to identify high-consumption activities

  2. Optimize your workflows using the tips in this guide

  3. Consider upgrading your plan if your usage consistently exceeds your allocation

  4. Set up low credit alerts in your account settings

Ready to optimize your credit usage? Start by reviewing your consumption patterns in Account Settings → Credit usage, then apply the optimization strategies from this guide to reduce unnecessary spending.