Introduction
If you've ever looked at an AI API pricing page and wondered what "per million tokens" actually means — you're not alone. Token pricing is the standard way AI companies charge for their APIs, but it can be confusing at first.
In this guide, we explain exactly what tokens are, how they're counted, and how to estimate your costs before you build.
What is a Token?
A token is a chunk of text — roughly 3-4 characters or about ¾ of a word in English.
Some examples:
- "Hello" = 1 token
- "Hello world" = 2 tokens
- "ChatGPT is amazing" = 4 tokens
- 1,000 words ≈ 750 tokens
- 1 million tokens ≈ 750,000 words (about 10 full novels)
Tokens aren't the same as words or characters — they depend on the model's tokenizer. Common words are usually 1 token, while long or rare words may be 2-3 tokens.
Input vs Output Tokens
Every AI API charges separately for input and output tokens:
- Input tokens: the text you send to the model — your prompt, system instructions, conversation history, and any context
- Output tokens: the text the model generates in response
Output tokens are almost always more expensive than input tokens. For example, GPT-4o charges $2.50/1M input but $10.00/1M output — 4x more expensive for output.
Current Token Prices (June 2026)
Premium models:
- GPT-4o: $2.50 input / $10.00 output per 1M tokens
- Claude Sonnet 4.6: $3.00 input / $15.00 output per 1M tokens
- Gemini 3.1 Pro: $2.00 input / $12.00 output per 1M tokens
Budget models:
- GPT-4o Mini: $0.15 input / $0.60 output per 1M tokens
- Claude Haiku 4.5: $1.00 input / $5.00 output per 1M tokens
- Gemini 3.1 Flash: $0.30 input / $2.50 output per 1M tokens
How to Calculate Your Costs
The formula is simple:
Cost = (Input tokens × Input price + Output tokens × Output price) × Number of requests
Example:
- You have 1,000 users
- Each sends 1 message per day (30/month)
- Average message: 200 input tokens + 300 output tokens
- Using GPT-4o Mini ($0.15 input / $0.60 output)
Monthly cost = (200 × $0.00000015 + 300 × $0.0000006) × 30,000 requests = ($0.00003 + $0.00018) × 30,000 = $6.30/month
What Increases Your Token Usage?
Several things silently add tokens to every request:
- System prompt: sent on every single request — keep it short
- Conversation history: sending past messages adds tokens fast
- RAG context: adding documents can add thousands of tokens per request
- Long responses: complex tasks generate more output tokens
Tips to Reduce Token Costs
- Use a budget model for simple tasks
- Keep your system prompt under 200 tokens
- Limit conversation history to last 5 messages
- Set a
max_tokenslimit on responses - Use context caching for repeated content
Use Our Free Calculator
Not sure how much your app will cost? Try our AI API Cost Calculator — enter your users, frequency, and use case to get an instant cost breakdown across all major models.
Conclusion
Token pricing is straightforward once you understand it. The key is knowing that input and output are priced separately, output costs more, and small optimizations to your prompts and context can save significant money at scale.