AI Prompt Cost Comparator
Enter how many input and output tokens a request uses and how many requests you expect, and see the cost across current model price presets side by side, cheapest first. Add your own models and prices for anything not listed, such as negotiated rates or a new release.
Result
Estimate
Lowest total: gpt-6-luna, 0.45 USD · Highest total: gpt-6-astra, 45.00 USD · Spread: 100×
| Model | Input price | Output price | Cost per request | Total cost | Compared with the lowest |
|---|---|---|---|---|---|
| gpt-6-luna OpenAI | 0.1 | 0.5 | 0.00045 USD | 0.45 USD | Lowest |
| Gemini 2.5 Flash GoogleGoogle currently limits access to the Gemini 2.5 models to users who have used them before. The model is not deprecated and no shutdown date is announced. Google’s model page | 0.3 | 2.5 | 0.00185 USD | 1.85 USD | 4.11× (+1.40 USD) |
| gpt-5.4-mini OpenAI | 0.75 | 4.5 | 0.00375 USD | 3.75 USD | 8.33× (+3.30 USD) |
| Claude Haiku 4.5 Anthropic | 1 | 5 | 0.0045 USD | 4.50 USD | 10× (+4.05 USD) |
| Claude Sonnet 5.5 Anthropic | 2 | 10 | 0.009 USD | 9.00 USD | 20× (+8.55 USD) |
| gpt-6.1-sol OpenAI | 2 | 10 | 0.009 USD | 9.00 USD | 20× (+8.55 USD) |
| Gemini 3.1 Pro Preview GooglePrice for prompts up to 200,000 tokens; longer prompts cost more. | 2 | 12 | 0.01 USD | 10.00 USD | 22.22× (+9.55 USD) |
| Claude Opus 5.5 Anthropic | 4 | 20 | 0.018 USD | 18.00 USD | 40× (+17.55 USD) |
| Claude Fable 5.1 Anthropic | 10 | 50 | 0.045 USD | 45.00 USD | 100× (+44.55 USD) |
| gpt-6-astra OpenAI | 10 | 50 | 0.045 USD | 45.00 USD | 100× (+44.55 USD) |
Preset prices are standard pay-as-you-go API prices in USD per million tokens, read from each provider’s pricing page and last verified 2026-10-19. Providers change prices often: check the source and use your own row when in doubt.
Price sources:
- Anthropic: Anthropic pricing page
- OpenAI: OpenAI API pricing page
- Google: Gemini API pricing page
Examples
Worked examples using the presets above (no caching):
| Workload | Input tokens | Output tokens | Requests | Lowest total | Highest total |
|---|---|---|---|---|---|
| Chat replies | 1,000 | 300 | 10,000 | gpt-6-luna: 2.50 USD | gpt-6-astra: 250.00 USD |
| Questions over retrieved documents | 8,000 | 500 | 1,000 | gpt-6-luna: 1.05 USD | gpt-6-astra: 105.00 USD |
| Summarising long documents | 50,000 | 1,000 | 100 | gpt-6-luna: 0.55 USD | gpt-6-astra: 55.00 USD |
Formula & Methodology
Given your token counts and the prices, the arithmetic is exact:
cost per request = input tokens × input price ÷ 1,000,000 + output tokens × output price ÷ 1,000,000with caching: cached share of the input uses the cached-input price insteadtotal cost = cost per request × number of requestsGood to know
- The same text is not the same number of tokens everywhere: every provider has its own tokenizer, and Anthropic documents about 30% more tokens for its newer models. Compare with each provider’s own token count when accuracy matters.
- Standard pay-as-you-go prices only. Batch, flex and priority tiers, cache-write and storage fees, long-prompt tiers, tool or search charges and taxes are not included.
- The cached-input price is the price for reading from the cache.
- The cheapest model is not automatically the best value: quality, speed, limits and reliability differ.
- Preset prices are conveniences, dated and sourced. Use your own rows for negotiated prices or new models.
How to Use
- Enter the input and output tokens of a typical request and the number of requests.
- Optionally enter the share of input served from cache.
- Add your own models and prices if something is missing.
- Read the table: it is sorted from the lowest total cost and shows how much more each option costs.
Frequently Asked Questions
How is the cost of an AI request calculated?
Providers charge per million tokens, with one price for input and a higher one for output. The cost is input tokens × input price plus output tokens × output price, each divided by one million, multiplied by the number of requests.
Are the preset prices up to date?
They were read from each provider’s pricing page on the date shown and linked as sources. Prices and model names change often, so check the source before relying on them and use a custom row for anything newer.
Why does the cheapest model not always win?
Price per token ignores quality, speed and how many tokens the model needs to get a good answer. A cheaper model that needs retries or longer prompts can cost more overall.
Why do different providers count different token numbers for the same text?
Each provider uses its own tokenizer, so the same text produces different counts. Anthropic, for example, documents that its newer models produce about 30% more tokens than earlier ones for the same text.
Can I compare a model that is not listed?
Yes. Fill in a row under “Your own models” with its name and its input, output and optional cached-input prices per million tokens, and it joins the comparison.
Last updated: