How Much Does AI Cost for a Small Business

How the Credit System Works

AI Apps API uses a credit-based pricing model. You buy credits and spend them as you use tools. One dollar buys 1,000 credits. Credits are consumed per operation: each chatbot response, each SMS sent, each email delivered, each workflow step executed. This means you never pay for capacity you do not use, and costs scale naturally with your actual usage.

If you bring your own API keys for AI providers (OpenAI, Anthropic), you pay the provider directly at their rates and skip the AI model markup on credits. This is an option for businesses with higher volume who want to optimize costs.

Cost Per Operation

AI Chatbot

• GPT-4.1-mini (recommended for support): 2-4 credits per response ($0.002-0.004)
• GPT-4.1 (more capable): 5-10 credits per response ($0.005-0.01)
• Claude Sonnet (balanced): 4-8 credits per response ($0.004-0.008)
• Reasoning models (complex tasks): 10-30 credits per response ($0.01-0.03)
• Knowledge base training: 3 credits per text chunk for embedding ($0.003)

A typical small business chatbot handling 50 conversations per day with GPT-4.1-mini uses about 150-200 credits daily, or roughly 4,500-6,000 credits per month ($4.50-6.00).

SMS Marketing

• Platform fee: 1-2 credits per message ($0.001-0.002)
• Carrier cost: Varies by provider, typically $0.005-0.02 per message (passed through from your SMS provider)
• AI-generated responses: Additional 2-4 credits if using AI to reply to incoming messages

Sending 500 messages per week costs about 500-1,000 platform credits ($0.50-1.00) plus carrier fees. The carrier fee is the larger cost component for SMS.

Email Marketing

• Platform fee: 1-2 credits per email ($0.001-0.002)
• SMTP provider cost: Varies, typically $0.0001-0.001 per email through services like Amazon SES, SendGrid, or Mailgun

Sending 2,000 emails per week costs about 2,000-4,000 platform credits ($2.00-4.00) plus minimal SMTP fees. Email is the cheapest channel per message.

Workflow Automation

• Non-AI steps (data routing, conditions, API calls): 1-3 credits per step
• AI-powered steps (content generation, decision making): 2-15 credits depending on AI model used

A simple workflow that triggers an SMS, updates a database, and sends a notification costs about 5-8 credits per execution.

Lead Generation

• Form submission processing: 1-2 credits
• Lead routing to buyers: 1-3 credits per route
• AI lead qualification: 3-8 credits per lead (depends on AI model)

Monthly Cost Estimates by Business Type

Small Service Business (salon, dental office, repair shop)

Typical tools: chatbot for questions, SMS reminders for appointments.

• Chatbot: 30 conversations/day x 3 credits = 2,700 credits/month ($2.70)
• SMS reminders: 200 messages/week x 2 credits = 1,600 credits/month ($1.60)
• Total platform credits: ~$4-5/month plus SMS carrier fees

Retail or Restaurant

Typical tools: chatbot, weekly SMS promotions, email newsletter.

• Chatbot: 40 conversations/day x 3 credits = 3,600 credits/month ($3.60)
• SMS campaigns: 1,000 messages/week x 2 credits = 8,000 credits/month ($8.00)
• Email: 2,000 emails/week x 1 credit = 8,000 credits/month ($8.00)
• Total platform credits: ~$20/month plus carrier and SMTP fees

Professional Services (real estate, insurance, consulting)

Typical tools: chatbot for lead capture, automated follow-up sequences, lead scoring.

• Chatbot: 20 conversations/day x 5 credits = 3,000 credits/month ($3.00)
• Lead follow-up drip: 300 messages/week x 2 credits = 2,400 credits/month ($2.40)
• AI lead qualification: 50 leads/week x 5 credits = 1,000 credits/month ($1.00)
• Workflow automation: 200 executions/week x 5 credits = 4,000 credits/month ($4.00)
• Total platform credits: ~$10-12/month plus carrier fees

How to Keep Costs Low

• Choose the right AI model. GPT-4.1-mini handles most chatbot conversations well at 2-4 credits. Only use more expensive models for tasks that actually require them. See When to Use a Cheap Model vs an Expensive One.
• Write good training data. A chatbot with well-organized, clear training data gives better answers on cheaper models. Bad training data forces you to use expensive models to compensate. See How to Organize Training Data for Best Results.
• Use your own API keys. If you spend more than $30/month on AI model credits, bringing your own OpenAI or Anthropic API key saves the platform markup.
• Monitor usage. Check your credit usage weekly to understand which tools consume the most and whether the cost matches the value.

Getting Started Without Risk

Start with a small credit purchase ($5-10) and test one tool. You will see exactly how fast credits are consumed and can project your monthly cost before committing further. There are no contracts, minimums, or cancellation fees. If you stop using the tools, you stop spending credits. See How Small Businesses Can Start Using AI Today for a step-by-step first-week plan.