# AI cost calculator.

URL: https://levystreet.com/resources/ai-cost-calculator/
Updated: 2026-10-19

AI cost calculator for NZ businesses: estimate LLM API running costs in NZD for 11 OpenAI, Anthropic and Google models, with caching, batch and exchange rate.

## Short answer

An AI cost calculator estimates what a language model costs to run by multiplying requests by input and output tokens and each model's per-token price. This calculator uses USD list prices for 11 OpenAI, Anthropic and Google models as listed on 19 October 2026, applies caching and batch discounts, and converts to New Zealand dollars at a rate you can change.

- Prices are USD per million tokens, as listed on 19 October 2026.
- The same workload ranges from about NZ$7 to NZ$694 a month across these models.
- Output costs five or more times the input price on these models, so long answers drive cost.
- Estimates exclude GST, cache writes, long-context surcharges and tool fees.

Use this calculator for anything that calls a model through an API: a document extraction pipeline, a support assistant, an agent or an internal tool. For per-seat subscriptions such as ChatGPT Business or Microsoft Copilot, see [AI tools pricing for New Zealand businesses](/resources/ai-tools-pricing-nz/) instead. Everything runs in your browser.

*[Interactive tool: llm-cost-calculator, available on the web page]*

## What each input means

| Input | What to enter | Default |
|---|---|---|
| Model | The model you plan to use. Prices are USD list prices as listed on 19 October 2026 | GPT-6.1 Sol |
| Requests per month | Number of model calls, not tasks. An agent may make several calls per task | 5,000 |
| Input tokens per request | Everything sent to the model: instructions, documents, conversation history, tool results | 6,000 |
| Output tokens per request | What the model writes back, including any reasoning the provider bills | 800 |
| Share of input served from cache (%) | The part of each prompt that repeats exactly across requests and is cached | 40% |
| USD to NZD rate | NZ dollars per US dollar. The ECB reference rates on 16 October 2026 give 1.7656 | 1.77 |
| Batch API | Tick if results can wait up to 24 hours; halves the price | Off |

The default rate of 1.77 is rounded from the European Central Bank's euro reference rates for 16 October 2026: US$1.1253 and NZ$1.9868 per euro, which gives NZ$1.7656 per US dollar ([ECB](https://www.ecb.europa.eu/stats/policy_and_exchange_rates/euro_reference_exchange_rates/html/index.en.html)). Your bank or card provider will add its own margin.

## The prices behind the calculator

USD per million tokens, standard tier, as listed on 19 October 2026 by [OpenAI](https://developers.openai.com/api/docs/pricing), [Anthropic](https://platform.claude.com/docs/en/about-claude/pricing) and [Google](https://ai.google.dev/gemini-api/docs/pricing).

| Model | Input | Cached input | Output | Notes |
|---|---|---|---|---|
| OpenAI GPT-6 Astra | 10.00 | 1.00 | 50.00 | Prompts over 272K tokens cost more |
| OpenAI GPT-6.1 Sol | 2.00 | 0.10 | 10.00 | Prompts over 272K tokens cost more |
| OpenAI GPT-6 Luna | 0.10 | 0.01 | 0.50 | |
| Anthropic Claude Fable 5.1 | 10.00 | 0.25 | 50.00 | |
| Anthropic Claude Opus 5.5 | 4.00 | 0.20 | 20.00 | |
| Anthropic Claude Sonnet 5.5 | 2.00 | 0.10 | 10.00 | |
| Anthropic Claude Haiku 5.5 | 0.10 | 0.01 | 0.50 | Prompts up to 100K tokens; higher above |
| Google Gemini 4 Argon, until 30 Nov 2026 | 2.00 | 0.10 | 10.00 | Introductory price |
| Google Gemini 4 Argon, from 1 Dec 2026 | 4.00 | 0.20 | 20.00 | Listed price after the introductory period |
| Google Gemini 3.8 Flash, until 31 Dec 2026 | 0.75 | 0.075 | 3.75 | Rises to 1.50 / 0.15 / 7.50 from 1 Jan 2027 |
| Google Gemini 3.5 Flash-Lite | 0.30 | 0.03 | 2.50 | |

## How the calculation works

In plain words: the input that is not cached is charged at the input price, the cached input at the cached price, and the output at the output price. Add them, divide by a million, halve it for batch, then convert to NZ dollars.

**Cost per request (NZ$)** = (input tokens × (1 − cache share) × input price + input tokens × cache share × cached price + output tokens × output price) ÷ 1,000,000 × 0.5 if batch × exchange rate.

Monthly cost is cost per request × requests per month; yearly is monthly × 12.

## Worked example

Illustrative example, using the defaults (GPT-6.1 Sol, 5,000 requests, 6,000 input and 800 output tokens, 40% cached, rate 1.77, no batch):

1. Uncached input: 3,600 tokens × US$2.00 per million = US$0.00720
2. Cached input: 2,400 tokens × US$0.10 per million = US$0.00024
3. Output: 800 tokens × US$10.00 per million = US$0.00800
4. Total per request: US$0.01544 × 1.77 = **NZ$0.027**
5. Per month: × 5,000 = **NZ$137**; per year **NZ$1,640**

Output is 52% of the cost from 12% of the tokens. The same workload on every model:

| Model | Per month | Per year |
|---|---|---|
| GPT-6 Luna or Claude Haiku 5.5 | NZ$7 | NZ$83 |
| Gemini 3.5 Flash-Lite | NZ$28 | NZ$335 |
| Gemini 3.8 Flash (until 31 Dec 2026) | NZ$52 | NZ$624 |
| GPT-6.1 Sol, Claude Sonnet 5.5 or Gemini 4 Argon (until 30 Nov) | NZ$137 | NZ$1,640 |
| Claude Opus 5.5 or Gemini 4 Argon (from 1 Dec) | NZ$273 | NZ$3,279 |
| Claude Fable 5.1 | NZ$678 | NZ$8,135 |
| GPT-6 Astra | NZ$694 | NZ$8,326 |

## Which inputs move the cost most

Starting from the GPT-6.1 Sol default of NZ$137 a month:

| Change | Monthly cost |
|---|---|
| Output 2,000 tokens instead of 800 | NZ$243 |
| No caching | NZ$177 |
| 80% of input cached | NZ$96 |
| Batch API | NZ$68 |
| Exchange rate 1.90 | NZ$147 |
| 50,000 requests instead of 5,000 | NZ$1,366 |

Model choice changes cost about a hundredfold; output length and volume come next. The cheapest model is only cheaper if it does the job: test candidates on your own cases first, as described in [AI model selection](/services/ai-model-selection/).

## Common mistakes and when the estimate is unreliable

- **Cache writes are excluded.** OpenAI charges 1.25 times the input price to write to the cache on GPT-5.6 and later models ([OpenAI](https://developers.openai.com/api/docs/guides/prompt-caching)), and Anthropic charges 1.25 times for five-minute cache writes.
- **Long prompts cost more.** OpenAI bills prompts over 272K input tokens at higher rates, and Claude Haiku 5.5 has a higher price above 100K tokens.
- **Agents make many calls.** One task can take several requests, each resending the conversation so far. Count calls, not tasks.
- **Tool fees are extra.** Web search costs US$10 per 1,000 calls on both OpenAI and Anthropic, plus tokens.
- **Token counts differ by provider.** Anthropic says Claude's newer tokenizer produces about 30% more tokens for the same text than older Claude models.
- **Prices change.** Gemini 4 Argon doubles on 1 December 2026 and Gemini 3.8 Flash on 1 January 2027.
- **GST and card fees are excluded.** From New Zealand, Anthropic's pricing page shows "Prices include 15% GST" beside its USD figures ([Anthropic](https://claude.com/pricing#api)); confirm on a real invoice.

## How to get real token counts


1. **Run 20 real requests.** Use your actual prompts and documents in a test account.
2. **Read the usage figures.** Every provider returns input, cached and output token counts with each response and in its console.
3. **Use the 90th percentile, not the average.** Long documents and long answers drive the bill.
4. **Count calls per task.** Especially for agents and multi-step workflows.
5. **Add headroom.** Retries, testing and growth usually add to the estimate; put the total into the [AI ROI calculator](/resources/ai-roi-calculator/) as part of monthly running cost.


## Next step

Our engineering guide to [managing LLM costs](/engineering/managing-llm-costs/) covers caching, routing to cheaper models and spending limits in production. If you are weighing per-token pricing against flat subscriptions, read [choosing between an API and a subscription](/learn/choosing-between-an-api-and-a-subscription/).

## Questions people ask

### How much does it cost to run an AI model for a business?

It depends on volume, how much text goes in and out of each request, and the model. With 5,000 requests a month, 6,000 input tokens and 800 output tokens each, and 40% of input cached, our calculator gives about NZ$7 a month on the cheapest current models and about NZ$690 on the most expensive, at USD list prices from 19 October 2026 before GST.

### What is a token?

A token is the unit models count text in, usually a short word or part of a word. Prices are quoted per million tokens. Token counts vary by provider for the same text; Anthropic says its newer tokenizer produces about 30% more tokens than older Claude models. Measure real token counts from your own requests rather than converting from word counts.

### Do NZ businesses pay GST on AI API usage?

The providers' API price lists are in US dollars before tax, and they do not set out New Zealand GST treatment clearly. Anthropic's pricing page shows 'Prices include 15% GST' when viewed from New Zealand, yet lists the same figures as its USD documentation. Check your actual invoice and ask your accountant how GST applies before budgeting.

### What is prompt caching and how much does it save?

Caching lets a provider reuse the unchanged start of a prompt, such as instructions or a long policy document, at a much lower price. On GPT-6.1 Sol, cached input costs 5% of the normal input price. It only helps when many requests share the same opening text, and some providers charge extra to write to the cache, which this calculator leaves out.

### When should we use the batch API?

When results can wait. Batch processing costs 50% less on all three providers, and OpenAI and Google target completion within 24 hours. It suits overnight jobs such as classifying a backlog of documents, enriching records or generating reports, not live chat or anything a customer is waiting for.


## Sources

- [API pricing](https://developers.openai.com/api/docs/pricing)
- [Prompt caching](https://developers.openai.com/api/docs/guides/prompt-caching)
- [Pricing](https://platform.claude.com/docs/en/about-claude/pricing)
- [Claude pricing (API section)](https://claude.com/pricing#api)
- [Gemini Developer API pricing](https://ai.google.dev/gemini-api/docs/pricing)
- [Euro foreign exchange reference rates](https://www.ecb.europa.eu/stats/policy_and_exchange_rates/euro_reference_exchange_rates/html/index.en.html)

Published by Levy Street (https://levystreet.com), Wellington, New Zealand.