LLM API cost calculator (OpenAI, Claude, Gemini)
AI APIs bill input and output tokens separately, per million tokens, and on most current models output costs four to five times as much as input. Enter the tokens per request and the requests per month to compare the monthly bill across OpenAI, Claude and Gemini models, with prompt caching and batch discounts.
GPT-6.1 Sol: 100,000 requests a month
$900.00
$0.009 a request · about ₹85,292 a month
- Per 1,000 requests
- $9.00
- Input price per 1M tokens
- $2.00standard
- Output price per 1M tokens
- $10.00
- Cheapest model for this workload
- OpenAI · GPT-4.1 nano$40.00 a month
| Model | Per request | Per month |
|---|---|---|
| OpenAI · GPT-4.1 nano | $0.0004 | $40.00 |
| Google · Gemini 2.5 Flash-Lite | $0.0004 | $40.00 |
| OpenAI · GPT-6 Luna | $0.00045 | $45.00 |
| Anthropic · Claude Haiku 5.5 | $0.00045 | $45.00 |
| OpenAI · GPT-4o Mini | $0.0006 | $60.00 |
| OpenAI · GPT-4.1 Mini | $0.0016 | $160.00 |
| Google · Gemini 3.5 Flash-Lite | $0.00185 | $185.00 |
| Google · Gemini 2.5 Flash | $0.00185 | $185.00 |
| Google · Gemini 3.8 Flash | $0.00337 | $337.50 |
| Google · Gemini 2.5 Pro | $0.0075 | $750.00 |
| OpenAI · GPT-4.1 | $0.008 | $800.00 |
| OpenAI · GPT-6.1 Sol | $0.009 | $900.00 |
| Anthropic · Claude Sonnet 5.5 | $0.009 | $900.00 |
| Google · Gemini 4 Argon | $0.009 | $900.00 |
| OpenAI · GPT-4o | $0.01 | $1,000.00 |
| Google · Gemini 3.1 Pro Preview | $0.01 | $1,000.00 |
| Anthropic · Claude Opus 5.5 | $0.02 | $1,800.00 |
| OpenAI · GPT-6 Astra | $0.05 | $4,500.00 |
| Anthropic · Claude Fable 5.1 | $0.05 | $4,500.00 |
- Prices checked 2026-10-20 from each vendor's pricing page; check them before you rely on these numbers.
Maps & AI Cost Audit: ₹45,000 / $950
At $900.00 a month this spend qualifies for the API cost audit's guarantee: savings worth 3× the fee over a year, or the fee back.
How is the cost of an AI API request calculated?
Every request is billed in tokens, in two parts at two prices: the input you send (system prompt, conversation, documents, images converted to tokens) and the output the model writes back. Prices are quoted per million tokens, so the cost of one request is input tokens × input price ÷ 1,000,000 plus output tokens × output price ÷ 1,000,000.
Two discounts change the answer. Input the vendor has already cached, such as a long system prompt repeated on every call, is billed at a fraction of the normal input price. The Batch API, for work that can wait up to a day, halves both input and output prices at all three vendors.
Some models switch to higher long-context rates for the whole request once the prompt passes a threshold, and some vendors have announced price changes or shutdown dates for older models. The calculator applies both and says so under the result.
cost per request = (fresh input × input price + cached input × cache price + output × output price) ÷ 1,000,000 monthly cost = cost per request × requests a month
Real token counts beat estimates. For the figures from apps we built, see our measured AI costs per scan, per card and per form.
Image inputs can cost far more tokens than a rule of thumb suggests; see what an AI plant scan costs.
The cheapest model is not always the cheapest result; read how we chose a vision model for card scanning.
How much do OpenAI, Claude and Gemini models cost per million tokens?
| Model | Input | Cached input | Output | Output ÷ input |
|---|---|---|---|---|
| OpenAI · GPT-6 Astra | $10.00 | $1.00 | $50.00 | 5× |
| OpenAI · GPT-6.1 Sol | $2.00 | $0.10 | $10.00 | 5× |
| OpenAI · GPT-6 Luna | $0.10 | $0.01 | $0.50 | 5× |
| OpenAI · GPT-4.1 | $2.00 | $0.50 | $8.00 | 4× |
| OpenAI · GPT-4.1 Mini | $0.40 | $0.10 | $1.60 | 4× |
| OpenAI · GPT-4.1 nano | $0.10 | $0.03 | $0.40 | 4× |
| OpenAI · GPT-4o | $2.50 | $1.25 | $10.00 | 4× |
| OpenAI · GPT-4o Mini | $0.15 | $0.08 | $0.60 | 4× |
| Anthropic · Claude Fable 5.1 | $10.00 | $0.25 | $50.00 | 5× |
| Anthropic · Claude Opus 5.5 | $4.00 | $0.20 | $20.00 | 5× |
| Anthropic · Claude Sonnet 5.5 | $2.00 | $0.10 | $10.00 | 5× |
| Anthropic · Claude Haiku 5.5 | $0.10 | $0.01 | $0.50 | 5× |
| Google · Gemini 4 Argon | $2.00 | $0.10 | $10.00 | 5× |
| Google · Gemini 3.8 Flash | $0.75 | $0.08 | $3.75 | 5× |
| Google · Gemini 3.5 Flash-Lite | $0.30 | $0.03 | $2.50 | 8.3× |
| Google · Gemini 3.1 Pro Preview | $2.00 | $0.20 | $12.00 | 6× |
| Google · Gemini 2.5 Pro | $1.25 | $0.13 | $10.00 | 8× |
| Google · Gemini 2.5 Flash | $0.30 | $0.03 | $2.50 | 8.3× |
| Google · Gemini 2.5 Flash-Lite | $0.10 | $0.01 | $0.40 | 4× |
How to estimate the monthly cost of an AI feature
- 1Pick a modelStart with the one you plan to use; the table under the result prices every other model on the same workload.
- 2Enter tokens per requestCount the prompt and the typical reply. Your vendor's dashboard shows real token counts once you have test traffic.
- 3Enter requests a month and the cache shareMultiply users by how often each one triggers the feature, and estimate how much of each prompt is the same every time.
- 4Compare and decideRead the monthly cost, then check whether a cheaper model or the Batch API would do the job.
Frequently asked questions
How many tokens is a word?
Why is output more expensive than input?
What does an AI feature cost per user in a real app?
Does prompt caching cut costs much?
When should I use the Batch API?
Sources
- OpenAI API pricing (checked 20 Oct 2026)
- OpenAI deprecations (checked 20 Oct 2026)
- Anthropic Claude pricing (checked 20 Oct 2026)
- Gemini Developer API pricing (checked 20 Oct 2026)
- FBIL USD/INR reference rate, 2026-10-19 (RBI website) (checked 20 Oct 2026)
Building a SaaS product?
We design, build, and scale SaaS platforms — from auth and billing to dashboards and APIs.
Built by CountryDevs · No signup · No ads · We count visits, never what you enter