How AI API pricing works
AI providers charge per token, a chunk of text that is usually part of a word. In English, 1,000 tokens is roughly 750 words. You pay separately for:
- Input tokens: everything you send, including your instructions, the conversation so far and any documents.
- Output tokens: everything the model writes back. These cost several times more than input tokens.
cost = (input tokens × input price + output tokens × output price) ÷ 1,000,000
Current prices
| Provider | Model | Input / 1M | Output / 1M | Notes |
|---|---|---|---|---|
| Anthropic | Claude Fable 5.2 | $8 | $40 | |
| Anthropic | Claude Opus 5.5 | $4 | $20 | |
| Anthropic | Claude Sonnet 5.5 | $2 | $10 | |
| Anthropic | Claude Haiku 5.5 | $0.10 | $0.50 | $0.50 / $2.50 above 100k input tokens |
| OpenAI | GPT-6.2 Astra | $8 | $40 | |
| OpenAI | GPT-6.2 Sol | $2 | $10 | |
| OpenAI | GPT-6.1 Terra | $1 | $5 | |
| OpenAI | GPT-6 Luna | $0.10 | $0.50 | |
| Gemini 3.1 Pro Preview | $2 | $12 | $4 / $18 above 200k input tokens | |
| Gemini 3.5 Flash | $1.50 | $9 | ||
| Gemini 3.1 Flash-Lite | $0.25 | $1.50 |
Prices change often, so always check the provider's own pricing page before you commit. Last checked 19 November 2026.
Five ways to cut your AI bill
- Use the smallest model that does the job. Small models are often 20 to 80 times cheaper and plenty for sorting, tagging and short replies.
- Keep prompts short. Long instructions are paid for on every single request.
- Cache repeated context. Most providers charge much less for input they have seen recently, such as a fixed system prompt.
- Batch work that can wait. Batch APIs typically cost about half the normal price, with results within a day.
- Limit the output length. Ask for short answers and set a maximum, since output tokens are the expensive part.