Pick up to three models, choose a usage preset or type your own numbers, and see what the same workload costs on each model per request, per day and per month, in US dollars and Indonesian rupiah.
With today's prices, a Claude Code-style workday (155 requests of about 151,000 input tokens, 65% cached, 650 output tokens) costs about $14.06 (Rp210,880) a month on deepseek-v4.1-flash-1017 and $14 (Rp210,045) on gpt-6-luna. A WhatsApp bot answering 1,000 chats a day costs about $4.58 (Rp68,748) a month on deepseek-v4.1-flash-1017, and an n8n workflow with 500 AI runs a day about $8.69 (Rp130,382).
155 agent steps a day (the busiest quarter of coding-agent users; the median is 45). Size of an average Claude Code request measured on Ezkielyna API: 151,233 input tokens, 65% of them cached, 650 output tokens.
Same usage, live prices per model. USD is the API balance used; rupiah at the price of $1 of API Credit.
Prices are the gateway's live per-model prices (https://api.ezkielyna.store/api/pricing), refreshed every hour: USD per 1M tokens = model ratio × 2 × the key-group rate (Standard 1×, Premium 1.5×). Output and cached input use each model's own multipliers.
Rupiah = USD × Rp15,000 per $1, the price of the $1 API Credit pack. Larger packs cost less per dollar (the $50 pack works out to Rp13,300 per $1), so real rupiah costs can be lower.
Cost per request = (uncached input × input price + cached input × cached price + output × output price) ÷ 1,000,000. A month is 30 days.
Reasoning (thinking) tokens are billed as output. Models that think a lot produce more output tokens than the answer you see, so raise the output number for them.
Caching is not guaranteed: it only applies when the beginning of the prompt repeats exactly, and models without a cached price bill all input at the normal rate.
Requests that fail on Ezkielyna's side or upstream before any answer are refunded to the balance automatically under the Gagal = Gratis policy (limits apply, see /jaminan), so the calculator counts successful requests only.
Questions people ask
How much does an AI API cost per month?
You pay per token: input tokens (everything you send, including chat history and files) and output tokens (the answer) each have a price per 1 million tokens. Cost per request = input × input price + output × output price, divided by 1,000,000; multiply by requests per day and 30 days. With live prices, a WhatsApp bot with 1,000 chats a day on deepseek-v4.1-flash-1017 is about $4.58 (Rp68,748) a month.
How many tokens is one message or one page of text?
As a rule of thumb one token is about 4 characters of English text; Indonesian usually needs a few more tokens per word (about 1.5 per word). A page of text is roughly 500–800 tokens. The exact input, cached and output tokens of every request are shown in the Log menu at api.ezkielyna.store.
What are cached input tokens and why are they cheaper?
When the start of a prompt is identical to an earlier request (system prompt, file contents, chat history), many models bill that part as cached input at a much lower price. Coding agents get large cache shares because they resend the same context every step; on Ezkielyna API 65–80% of coding-agent input is billed as cache.
Why does Claude Code cost more per request than a chatbot?
A coding agent resends the system prompt, tool list, the files it read and every previous step in each request, so an average Claude Code request on Ezkielyna API carries about 151,000 input tokens against about 1,300 for a short chat message. Caching and a cheaper model for simple steps keep the bill down.
Can I pay for API usage in rupiah?
Yes. Ezkielyna API Credit is bought with QRIS (DANA, GoPay, OVO, ShopeePay, mobile banking), USDT or Binance Pay, redeemed in Wallet at api.ezkielyna.store, and charged per token. The balance does not expire. This calculator converts at Rp15,000 per $1, the price of the $1 pack.
Ringkasan dalam Bahasa Indonesia
Kalkulator biaya API AI gratis: bandingkan sampai 3 model sekaligus (DeepSeek, GLM, Kimi, Qwen, Claude, GPT dan lainnya) dengan harga live Ezkielyna API. Isi request per hari, token input dan output serta persentase cache, lalu lihat biaya per request, per hari dan per bulan dalam USD dan Rupiah. Ada contoh untuk Claude Code, bot WhatsApp 1.000 chat dan workflow n8n 500 run. Contoh hari ini: Claude Code 155 request per hari dengan deepseek-v4.1-flash-1017 sekitar $14.06 (Rp210.880) per bulan. Pilih "Bahasa Indonesia" di atas untuk tampilan lengkap.
