LLM cost calculator
Enter your monthly token volume and see the estimated monthly bill side by side across OpenAI, Anthropic, and open-source models on Gonka. Per-token rates are hard to feel; a monthly bill is not.
Enter the tokens you burn in a day if you know them, or pick the closest workload if you don't.
| Provider and model | Est. monthly cost |
|---|---|
| OpenAI GPT-6 Astra | $6,000 |
| OpenAI GPT-5.6 Terra | $1,350 |
| Anthropic Claude Fable 5.1 | $6,000 |
| Anthropic Claude Sonnet 5 | $1,200 |
| Gonka · MiniMax M2 | $75 |
| Gonka · DeepSeek V4 Flash | $60 |
| Gonka · GLM-5.3-Flash | $60 |
More about plans, credits, and limits: OpenAI API cost and Claude API cost.
Estimated comparison at 75% input / 25% output tokens, list prices as of September 2026. Prices and terms change over time; always check current rates.
How the numbers are computed
- Monthly cost = tokens per month × the blended per-token price at an assumed 75% input / 25% output mix, typical for automation: prompts carry context and documents in, answers are short.
- OpenAI and Anthropic prices are the published API list prices as of September 2026, taken from the official pricing pages: OpenAI, Anthropic. Prices change over time for every provider; always check current rates. Discounts like batch processing and prompt caching are not modeled.
- Gonka prices are our current per-token rates, the same for input and output, locked in when you top up. Details on the Gonka API pricing page.
For the full story behind these numbers, see the cost breakdowns: OpenAI API cost and Claude API cost.
Token calculator FAQ
How many tokens is a word?
For English text, one token is roughly three quarters of a word, so 1M tokens is about 750,000 words: around ten average novels. Code, JSON, and non-English text tokenize less efficiently, so real workloads often use more tokens than a word count suggests.
How do I estimate my monthly token volume?
Easiest through examples. Transcribing and summarizing one hour of a Zoom call runs roughly 15,000 tokens. A support bot handling 50 tickets a day burns about 100,000 a day. A coding agent working alongside one developer for a day can run into millions. A pipeline classifying 10,000 documents a day at about 1,000 tokens each is 10M a day, 300M a month. Pick the example closest to your case and scale it.
What about the $200 subscriptions?
At relatively modest, interactive volumes a $200 consumer plan gives you a flat monthly price within its limits, and that can beat any per-token rate. The calculator sticks to what can be computed honestly: vendors do not publish token limits for their plans. For products, SaaS, and volumes beyond plan limits, per-token prices are what counts; the cost breakdowns linked below the calculator cover the plans in detail.
Like what the math says?
Create an account, top up in USD, and your per-token price is locked in for the whole balance.
Get startedAlready have an account? Sign in